Allison Morrow, CNN:

Tech columnists such as the New York Times’ Kevin Roose have suggested recently that Apple has failed AI, rather than the other way around.

“Apple is not meeting the moment in AI,” Roose said on his podcast, Hard Fork, earlier this month. “I just think that when you’re building products with generative AI built into it, you do just need to be more comfortable with error, with mistakes, with things that are a little rough around the edges.”

To which I would counter, respectfully: Absolutely not.

Via Dan Moren, of Six Colors:

The thesis of the piece is not about excusing Apple’s AI missteps, but zooming out to take a look at the bigger picture of why AI is everywhere, and make the argument that maybe Apple is well-served by not necessarily being on the cutting edge of these developments.

If that is what this piece is arguing, I do not think Apple makes a good case for it. When it launched Apple Intelligence, it could have said it was being more methodical, framing a modest but reliable feature set as a picture of responsibility. This would be a thin layer of marketing speak covering the truth, of course, but that would at least set expectations. Instead, what we got was a modest and often unreliable feature set with mediocre implementation, and the promise of a significantly more ambitious future that has been kicked down the road.

These things do not carry the Apple promise, as articulated by Morrow, of “design[ing] things that are accessible out of the box”, products for which “[y]ou will almost never need a user manual filled with tiny print”. It all feels flaky and not particularly nice to use. Even the toggle to turn it off is broken.

⌥ Permalink

Meta:

Formerly a place to view friend requests and People You May Know, the Friends tab will now show your friends’ stories, reels, posts, birthdays and friend requests.

You know, I think this concept of showing people things they say they want to see might just work.

Meta says this is just one of “several ‘O.G.’ Facebook experiences [coming] throughout the year” — a truly embarrassing sentence. But Mark Zuckerberg said in an autumn earnings call that Facebook would “add a whole new category of content which is A.I. generated or A.I. summarized content, or existing content pulled together by A.I. in some way”. This plan is going just great. I think the way these things can be reconciled is exactly how Facebook is doing it: your friends go in a “Friends” tab, but you will see all the other stuff it wants to push on you by default. Just look how Meta has done effectively the same thing in Instagram and Threads.

⌥ Permalink

Reid Tatoris, Harsh Saxena, and Luis Miglietti, of Cloudflare:

Today, we’re excited to announce AI Labyrinth, a new mitigation approach that uses AI-generated content to slow down, confuse, and waste the resources of AI Crawlers and other bots that don’t respect “no crawl” directives. When you opt in, Cloudflare will automatically deploy an AI-generated set of linked pages when we detect inappropriate bot activity, without the need for customers to create any custom rules.

Two thoughts:

1. This is amusing. Nothing funnier than using someone’s own words or, in this case, technology against them.

2. This is surely going to lead to the same arms race as exists now between privacy protections and hostile adtech firms. Right?

⌥ Permalink

There is a long line of articles questioning Apple’s ability to deliver on artificial intelligence because of its position on data privacy. Today, we got another in the form of a newsletter.

Reed Albergotti, Semafor:

Meanwhile, Apple was focused on vertically integrating, designing its own chips, modems, and other components to improve iPhone margins. It was using machine learning on small-scale projects, like improving its camera algorithms.

[…]

Without their ads businesses, companies like Google and Meta wouldn’t have built the ecosystems and cultures required to make them AI powerhouses, and that environment changed the way their CEOs saw the world.

Again, I will emphasize this is a newsletter. It may seem like an article from a prestige publisher that prides itself on “separat[ing] the facts from our views”, but you might notice how, aside from citing some quotes and linking to ads, none of Albergotti’s substantive claims are sourced. This is just riffing.

I remain skeptical. Albergotti frames this as both a mindset shift and a necessity for advertising companies like Google and Meta. But the company synonymous with the A.I. boom, OpenAI, does not have the same business model. Besides, Apple behaves like other A.I. firms by scraping the web and training models on massive amounts of data. The evidence for this theory seems pretty thin to me.

But perhaps a reluctance to be invasive and creepy is one reason why personalized Siri features have been delayed. I hope Apple does not begin to mimic its peers in this regard; privacy should not be sacrificed. I think it is silly to be dependent on corporate choices rather than legislation to determine this, but that is the world some of us live in.

Let us concede the point anyhow, since it suggests a role Apple could fill by providing an architecture for third-party A.I. on its products. It does not need to deliver everything to end users; it can focus on building a great platform. Albergotti might sneeze at “designing its own chips […] to improve iPhone margins”, which I am sure was one goal, but it has paid off in ridiculously powerful Macs perfect for A.I. workflows. And, besides, it has already built some kind of plugin architecture into Apple Intelligence because it has integrated ChatGPT. There is no way for other providers to add their own extension — not yet, anyhow — but the system is there.

Gus Mueller:

The crux of the issue in my mind is this: Apple has a lot of good ideas, but they don’t have a monopoly on them. I would like some other folks to come in and try their ideas out. I would like things to advance at the pace of the industry, and not Apple’s. Maybe with a blessed system in place, Apple could watch and see how people use LLMs and other generative models (instead of giving us Genmoji that look like something Fisher-Price would make). And maybe open up the existing Apple-only models to developers. There are locally installed image processing models that I would love to take advantage of in my apps.

Via Federico Viticci, MacStories:

Which brings me to my second point. The other feature that I could see Apple market for a “ChatGPT/Claude via Apple Intelligence” developer package is privacy and data retention policies. I hear from so many developers these days who, beyond pricing alone, are hesitant toward integrating third-party AI providers into their apps because they don’t trust their data and privacy policies, or perhaps are not at ease with U.S.-based servers powering the popular AI companies these days. It’s a legitimate concern that results in lots of potentially good app ideas being left on the table.

One of Apple’s specialties is in improving the experience of using many of the same technologies as everyone else. I would like to see that in A.I., too, but I have been disappointed by its lacklustre efforts so far. Even long-running projects where it has had time to learn and grow have not paid off, as anyone can see in Siri’s legacy.

What if you could replace these features? What if Apple’s operating systems were great platforms by which users could try third-party A.I. services and find the ones that fit them best? What if Apple could provide certain privacy promises, too? I bet users would want to try alternatives in a heartbeat. Apple ought to welcome the challenge.

An odd little detail in a Physics World piece on Microsoft’s claim to have made a working topological qubit:

Regardless of the debate about the results and how they have been announced, researchers are supportive of the efforts at Microsoft to produce a topological quantum computer. “As a scientist who likes to see things tried, I’m grateful that at least one player stuck with the topological approach even when it ended up being a long, painful slog,” says [Scott] Aaronson.

“Most governments won’t fund such work, because it’s way too risky and expensive,” adds [Winfried] Hensinger. “So it’s very nice to see that Microsoft is stepping in there.”

In drug development, defence technologies, and life sciences research, to name a few, we’ve seen the opposite: governments fund the risky, expensive part for many years, often decades, until something viable emerges. Then the IP moves to public and private sector enterprises for commercialisation, sometimes together with government subsidies to increase public access. With pharmaceuticals in particular, the government often doesn’t recoup investments it has made in the discovery phase, which includes medical education and research. An illustrative recent example is the development of mRNA vaccines; from my piece in The Hinducriticising the medicine Nobel Prize for this work:

Dr. Kariko and Dr. Weissman began working together on the mRNA platform at the University of Pennsylvania in the late 1990s. The University licensed its patents to mRNA RiboTherapeutics, which sublicensed them to CellScript, which sublicensed them to Moderna and BioNTech for $75 million each. Dr. Karikó joined BioNTech as senior vice-president in 2013, and the company enlisted Pfizer to develop its mRNA vaccine for COVID-19 in 2020.

Much of the knowledge that underpins most new drugs and vaccines is unearthed at the expense of governments and public funds. This part of drug development is more risky and protracted, when scientists identify potential biomolecular targets within the body on which a drug could act in order to manage a particular disease, followed by identifying suitable chemical candidates. The cost and time estimates of this phase are $1billion-$2.5 billion and several decades, respectively.

Companies subsequently commoditise and commercialise these entities, raking in millions in profits, typically at the expense of the same people whose taxes funded the fundamental research. There is something to be said for this model of drug and vaccine development, particularly for the innovation it fosters and the eventual competition that lowers prices, but we cannot deny the ‘double-spend’ it imposes on consumers — including governments — and the profit-seeking attitude it engenders among the companies developing and manufacturing the product.

Quantum computing may well define the next technological revolution together with more mature AI models. Topological quantum computing in particular — if realised well enough to compete with alternative architectures based on superconducting wires and/or trapped ions — could prove especially valuable for its ability to be more powerful with fewer resources. Governments justify their continuing sizeable expense on drug development by the benefits that eventually accrue to the country’s people. By all means, quantum technologies will have similar consequences, following from a comparable trajectory of development where certain lines of inquiry are not precluded because they could be loss-making or amount to false starts. And they will impinge on everything from one’s fundamental rights to national security.

But Hensinger’s opinion indicates the responsibility of developing this technology has been left to the private sector. I wonder if there are confounding factors here. For example, is Microsoft’s pursuit of a topological qubit the exception to the rule — i.e. one of a few enterprises that are funded by a private organisation in a sea of publicly funded research? Another possibility is that we’re hearing about Microsoft’s success because it has a loud voice, with the added possibility that its announcement was premature (context here). It’s also possible Microsoft’s effort included grants from NSF, DARPA or the like.

All this said, let’s assume for a moment that what Hensinger said was true of quantum computing research in general: the lack of state-led development in such potentially transformative technologies raises two (closely related) concerns. The first is scientific progress, especially that it will happen behind closed doors. In a June 2023 note, senior editors of the Physical Review B journal acknowledged the contest between the importance of researchers sharing their data for scrutiny, replication, and for others to build on their work — all crucial for science — and private sector enterprises’ need to protect IP and thus withhold data. “This will not be the last time the American Physical Society confronts a tension between transparency and the transmission of new results,” they added. Unlike in drug development, life sciences, etc., even the moral argument that publicly funded research must be in the public domain is rendered impotent, although it can still be recast as the weaker “research that affects the public sphere…”.

The second is democracy. In a March 2024 commentary, digital governance experts Nathan Sanders, Bruce Schneier, and Norman Eisen wrote that the state could develop a “public AI” to counter the already apparent effects of “private AI” on democratic institutions. According to them, a “public AI” model could “provide a mechanism for public input and oversight on the critical ethical questions facing AI development,” including “how to incorporate copyrighted works in model training” and “how to license access for sensitive applications ranging from policing to medical use”. They added: “Federally funded foundation AI models would be provided as a public service, similar to a health care private option. They would not eliminate opportunities for private foundation models, but they would offer a baseline of price, quality, and ethical development practices that corporate players would have to match or exceed to compete.”

Of course, quantum computing isn’t beset by the same black-box problem that surrounds AI models, yet what it implies for our ability to secure digital data means it could still benefit from state-led development. Specifically: (i) a government-funded technology standard could specify the baseline for the private sector to “match or exceed to compete” so that computers deployed to secure public data maintain a minimum level of security; (ii) private innovation can build on the standard, with the advantage of not having to lay new foundations of their own; and (iii) the data and the schematics pertaining to the standard should be in the public domain, thus restricting private-sector IP to specific innovations.[1]

[1] Contrary to a lamentable public perception, just knowing how a digital technology works doesn’t mean it can be hacked.

I have access to the premium version of ChatGPT, and every day I ask it a few questions about concepts in physics that I’d like to know more about. Yesterday, for example, I learnt the difference between quenching and annealing…

Is there an opposite phenomenon for quenching?

Yes! The opposite of quenching would be processes that allow gradual relaxation, equilibration, or annealing rather than rapid freezing. Some terms that describe this include:

Annealing (Thermal and Simulated) — In metallurgy and physics, annealing is the process of slow cooling a material so it can reach a more ordered, lower-energy state.

In simulated annealing (used in optimization and machine learning), a system is gradually "cooled" to allow it to find a global minimum rather than getting trapped in local minima (which can happen in quenching).

Slow Equilibration / Adiabatic Processes — Unlike quenching, which often leads to metastable states, slow thermalization allows the system to reach true equilibrium.

Adiabatic processes in quantum mechanics and condensed matter physics allow the system to evolve slowly so that it remains in a well-defined state.

Supercooling vs. Slow Nucleation — When liquids are rapidly cooled (quenched), they might bypass crystallization and form a glass. In contrast, if cooling is slow, proper nucleation occurs, leading to a well-ordered crystal.

Self-Organization & Aging in Complex Systems — Some complex systems exhibit aging (e.g., glasses, biological systems) where they slowly evolve toward an ordered or more stable configuration over time, in contrast to being frozen in a quenched state.

… and later about the Weierstrass function.

Please explain the Weierstrass function to me.

The Weierstrass function is a famous example of a continuous but nowhere differentiable function. It is defined as:

[ W(x) = \sum_{n=0}^{\infty} aⁿ \cos(bⁿ \pi x) ]

where:

• ( 0 < a < 1 ) (controls decay)
• ( b ) is an odd integer ( > 1 ) (controls frequency)
• The product ( ab > 1 + \frac{3\pi}{2} ) ensures nowhere differentiability.

Key Properties

Continuous Everywhere — Despite its wild oscillations, ( W(x) ) is continuous for all ( x ).

Nowhere Differentiable — No well-defined slope exists at any point, making it an early example of a function that is continuous yet "rough" at all scales.

Self-Similarity (Fractal-like Behavior) — Since it is constructed as an infinite sum of cosine waves with exponentially increasing frequencies, it exhibits self-similarity: small-scale structures resemble larger-scale patterns.

Role in Mathematics — This function was introduced by Karl Weierstrass in the 19th century as a counterexample to the assumption that "continuous functions are differentiable almost everywhere."

I have found using ChatGPT in this way to be a valuable learning tool, and I imagine there are already people out there who are repackaging instances of ChatGPT to be autonomous ‘teachers’ for students looking to learn something.

One of my jobs at The Hindu is to get explanatory articles for concepts in (pure) science. I recently attempted one on amplifiers, which required me to explain the working of a NPN bipolar junction transistor, a device I’d until then been happy to leave behind in my high school classroom. I turned to ChatGPT, asking for a visualisation of the transistor, and it obliged. I’m a visual learner and having access to the tool made a big difference.

How does an NPN bipolar junction transistor work? Please explain with a visualisation.

An NPN bipolar junction transistor (BJT) is a type of transistor that consists of three layers of semiconductor material: two n-type layers (emitter and collector) separated by a p-type layer (base). It is commonly used for switching and amplification.

How an NPN Transistor Works

Structure & Terminals:

Emitter (E): Heavily doped n-type region that emits electrons.

Base (B): Thin, lightly doped p-type region.

Collector (C): Moderately doped n-type region that collects electrons.

Operation Principles:

When a small current is applied to the base-emitter junction (forward-biased), it allows a larger current to flow from the emitter to the collector.

The base-emitter junction acts as a diode, allowing electrons to flow into the base.

Most electrons injected from the emitter into the base are swept into the collector due to the voltage difference.

Visualization

      Collector (C)
           |
           |
      |----|----|
      |    N    |   <-- Collector (N-type)
      |----|----|
           |
      |----|----|
      |    P    |   <-- Base (P-type, thin and lightly doped)
      |----|----|
           |
      |----|----|
      |    N    |   <-- Emitter (N-type, heavily doped)
      |----|----|
           |
           |
      Emitter (E)

Current Flow:

Electrons flow from Emitter (E) to Collector (C).

A small Base (B) current controls a much larger Collector (C) current.

Voltage Conditions:

V_BE (Base-Emitter voltage) > 0.7V (Forward biased)

V_CE (Collector-Emitter voltage) > V_BE (Reverse biased)

Electron Flow:

→ Represents electron flow.

I_B (Base Current) is small but controls I_C (Collector Current), which is large.

( +V_C ) | | [C] ← Electrons | | | [B] → Small Base Current (I_B) | | | [E] → Large Electron Flow (I_E) | ( GND )

Working Mechanism:

The Base-Emitter junction is forward-biased, allowing electrons to move from Emitter to Base.

Most electrons are swept into the Collector due to the reverse-biased Collector-Base junction.

A small Base current (I_B) controls a large Collector current (I_C), amplifying the input signal.

I have a background in engineering plus more than a decade’s experience in spotting yellow/red flags in scientific papers, and I imagine anyone with these skills will have an easier time navigating ChatGPT’s answers. For anyone else that lacks these skills, all we need is a guarantee from OpenAI that the tool doesn’t hallucinate or that it hallucinates in specific contexts, and definitely not above a certain rate.

Have you seen the new ads for Google Gemini?

In one version, just as a young employee is grabbing her fast-food lunch, she notices her snooty boss get on an elevator. So she drops her sandwich, rushes to meet her just as the doors are about to close, and submits her proposal in the form of a thick dossier. The boss asks her for a 500-word summary to consume during her minute-long elevator ride. The employee turns to Google Gemini, which digests the report and spits out the gist, and which the employee regurgitates to the boss’s approval. The end.

Isn’t this unsettling? Google isn’t alone either. In May this year, Apple released a tactless ad for its new iPad Pro. From Variety:

The “Crush!” ad shows various creative and cultural objects — including a TV, record player, piano, trumpet, guitar, cameras, a typewriter, books, paint cans and tubes, and an arcade game machine — getting demolished in an industrial press. At the end of the spot, the new iPad Pro pops out, shiny and new, with a voiceover that says, “The most powerful iPad ever is also the thinnest.”

After the backlash, Apple bactracked and apologised — and then produced two ads in November for its Apple Intelligence product showcasing how it could help thoughtless people continue to be thoughtless.

The second video is additionally weird because it seems to suggest reaching all the way for an AI tool makes more sense than setting a reminder on the calendar that comes in all smartphones these days.

And they are now joined in spirit by Google, because bosses can now expect their subordinates to Geminify their way through what could otherwise have been tedious work or just impossible to do on punishingly short deadlines — without the bosses having to think about whether their attitudes towards what they believe is reasonable to ask of their teammates need to change. (This includes a dossier of details that ultimately won’t be read.)

If AI is going to absorb the shock that comes of someone being crappy to you, will we continue to notice that crappiness and demand they change or — as Apple and Google now suggest — will we blame ourselves for not using AI to become crappy ourselves? To quote from a previous post:

When machines make decisions, the opportunity to consider the emotional input goes away. This is a recurring concern I’m hearing about from people working with or responding to AI in some way. … This is Anna Mae Duane, director of the University of Connecticut Humanities Institute, in The Conversation: “I fear how humans will be damaged by the moral vacuum created when their primary social contacts are designed solely to serve the emotional needs of the ‘user’.”

The applications of these AI tools have really blossomed and millions of people around the world are using them for all sorts of tasks. But even if the ads don’t pigeonhole these tools, they reveal how their makers — Apple and Google — are thinking about what the tools bring to the table and what these tech companies believe to be their value. To Google’s credit at least, its other ads in the same series are much better (see here and here for examples), but they do need to actively cut down on supporting or promoting the idea that crappy behaviour is okay.

X on Wednesday announced a new set of terms, something which is normally a boring and staid affair. But these are a doozy:

Here’s a high-level recap of the primary changes that go into effect on November 15, 2024. You may see an in-app notice about these updates as well.

• Governing law and forum changes: For users residing outside of the European Union, EFTA States, and the United Kingdom, we’ve updated the governing law and forum for lawsuits to Texas as specified in our terms. […]

Specifically, X says “disputes […] will be brought exclusively in the U.S. District Court for the Northern District of Texas or state courts located in Tarrant County, Texas, United States”. X’s legal address is on a plot of land shared with SpaceX and the Boring Company near Bastrop, which is in the Western District. This particular venue is notable as the federal judge handling current X litigation in the Northern District owns Tesla stock and has not recused himself in X’s suit against Media Matters, despite stepping aside on a similar case because of a much smaller investment in Unilever. The judge, Reed O’Connor, is a real piece of work from the Federalist Society who issues reliably conservative decisions and does not want that power undermined.

An investment in Tesla does not necessarily mean a conflict of interest with X, an ostensibly unrelated company — except it kind of does, right? This is the kind of thing the European Commission is trying to figure out: are all of these different businesses actually related because they share the same uniquely outspoken and influential figurehead? Musk occupies such a particularly central role in all these businesses and it is hard to disentangle him from their place in our society. O’Connor is not the only judge in the district, but it is notable the company is directing legal action to that venue.

But X is only too happy to sue you in any court of its choosing.

Another of the X terms updates:

• AI and machine learning clarifications: We’ve added language to our Privacy Policy to clarify how we may use the information you share to train artificial intelligence models, generative or otherwise.

This is rude. It is a “clarifi[cation]” described in vague terms, and what it means is that users will no longer be able to opt out of their data being used to train Grok or any other artificial intelligence product. This appears to also include images and video, posts in private accounts and, if I am reading this right, direct messages.

Notably, Grok is developed by xAI, which is a completely separate company from X. See above for how Musk’s companies all seem to bleed together.

• Updates to reflect how our products and services work: We’ve incorporated updates to better reflect how our existing and upcoming products, features, and services work.

I do not know what this means. There are few product-specific changes between the old and new agreements. There are lots — lots — of new ways X wants to say it is not responsible for anything at all. There is a whole chunk which effectively replicates the protections of Section 230 of the CDA, you now need written permission from X to transfer your account to someone else, and X now spells out its estimated damages from automated traffic: $15,000 USD per million posts every 24 hours.

Oh, yeah, and X is making block work worse:

If your posts are set to public, accounts you have blocked will be able to view them, but they will not be able to engage (like, reply, repost, etc.).

The block button is one of the most effective ways to improve one’s social media experience. From removing from your orbit people who you never want to hear from for even mundane reasons, to reducing the ability for someone to stalk or harass, its expected action is vital. This sucks. I bet the main reason this change was made is because Musk is blocked by a lot of people.

All of these changes seem designed to get rid of any remaining user who is not a true believer. Which brings us to today.

Sarah Perez, TechCrunch:

Social networking startup Bluesky, which just reported a gain of half a million users over the past day, has now soared into the top five apps on the U.S. App Store and has become the No. 2 app in the Social Networking category, up from No. 181 a week ago, according to data from app intelligence firm Appfigures. The growth is entirely organic, we understand, as Appfigures confirmed the company is not running any App Store Search Ads.

As of writing, Bluesky is the fifth most popular free app in the Canadian iOS App Store, and the second most popular free app in the Social Networking category. Threads is the second most popular free app, and the most popular in the Social Networking category.

X is number 74 on the top free apps list. It remains classified as “News” in the App Store because it, like Twitter, has always compared poorly against other social media apps.

⌥ Permalink

After Robb Knight found — and Wired confirmed — Perplexity summarizes websites which have followed its opt out instructions, I noticed a number of people making a similar claim: this is nothing but a big misunderstanding of the function of controls like robots.txt. A Hacker News comment thread contains several versions of these two arguments:

• robots.txt is only supposed to affect automated crawling of a website, not explicit retrieval of an individual page.

• It is fair to use a user agent string which does not disclose automated access because this request was not automated per se, as the user explicitly requested a particular page.

That is, publishers should expect the controls provided by Perplexity to apply only to its indexing bot, not a user-initiated page request. Wary of being the kind of person who replies to pseudonymous comments on Hacker News, this is an unnecessarily absolutist reading of how site owners expect the Robots Exclusion Protocol to work.

To be fair, that protocol was published in 1994, well before anyone had to worry about websites being used as fodder for large language model training. And, to be fairer still, it has never been formalized. A spec was only recently proposed in September 2022. It has so far been entirely voluntary, but the draft standard proposes a more rigid expectation that rules will be followed. Yet it does not differentiate between different types of crawlers — those for search, others for archival purposes, and ones which power the surveillance economy — and contains no mention of A.I. bots. Any non-human means of access is expected to comply.

The question seems to be whether what Perplexity is doing ought to be considered crawling. It is, after all, responding to a direct retrieval request from a user. This is subtly different from how a user might search Google for a URL, in which case they are asking whether that site is in the search engine’s existing index. Perplexity is ostensibly following real-time commands: go fetch this webpage and tell me about it.

But it clearly is also crawling in a more traditional sense. The New York Times and Wired both disallow PerplexityBot, yet I was able to ask it to summarize a set of recent stories from both publications. At the time of writing, the Wired summary is about seventeen hours outdated, and the Times summary is about two days old. Neither publication has changed its robots.txt directives recently; they were both blocking Perplexity last week, and they are blocking it today. Perplexity is not fetching these sites in real-time as a human or web browser would. It appears to be scraping sites which have explicitly said that is something they do not want.

Perplexity should be following those rules and it is shameful it is not. But what if you ask for a real-time summary of a particular page, as Knight did? Is that something which should be identifiable by a publisher as a request from Perplexity, or from the user?

The Robots Exclusion Protocol may be voluntary, but a more robust method is to block bots by detecting their user agent string. Instead of expecting visitors to abide by your “No Homers Club” sign, you are checking IDs. But these strings are unreliable and there are often good reasons for evading user agent sniffing.

Perplexity says its bot is identifiable by both its user agent and the IP addresses from which it operates. Remember: this whole controversy is that it sometimes discloses neither, making it impossible to differentiate Perplexity-originating traffic from a real human being — and there is a difference.

A webpage being rendered through a web browser is subject to the quirks and oddities of that particular environment — ad blockers, Reader mode, screen readers, user style sheets, and the like — but there is a standard. A webpage being rendered through Perplexity is actually being reinterpreted and modified. The original text of the page is transformed through automated means about which neither the reader or the publisher has any understanding.

This is true even if you ask it for a direct quote. I asked for a full paragraph of a recent article and it mashed together two separate sections. They are direct quotes, to be sure, but the article must have been interpreted to generate this excerpt.¹

It is simply not the case that requesting a webpage through Perplexity is akin to accessing the page via a web browser. It is more like automated traffic — even if it is being guided by a real person.

The existing mechanisms for restricting the use of bots on our websites are imperfect and limited. Yet they are the only tools we have right now to opt out of participating in A.I. services if that is something one wishes to do, short of putting pages or an entire site behind a user name and password. It is completely reasonable for someone to assume their signal of objection to any robotic traffic ought to be respected by legitimate businesses. The absolute least Perplexity can do is respecting those objections by clearly and consistently identifying itself, and excluding websites which have indicated they do not want to be accessed by these means.

Takeshi Narabe, the Asahi Shimbun:

SoftBank Corp. announced that it has developed voice-altering technology to protect employees from customer harassment.

The goal is to reduce the psychological burden on call center operators by changing the voices of complaining customers to calmer tones.

The company launched a study on “emotion canceling” three years ago, which uses AI voice-processing technology to change the voice of a person over a phone call.

Penny Crosman, the American Banker:

Call center agents who have to deal with angry or perplexed customers all day tend to have through-the-roof stress levels and a high turnover rate as a result. About 53% of U.S. contact center agents who describe their stress level at work as high say they will probably leave their organization within the next six months, according to CMP Research’s 2023-2024 Customer Contact Executive Benchmarking Report.

Some think this is a problem artificial intelligence can fix. A well-designed algorithm could detect the signs that a call center rep is losing it and do something about it, such as send the rep a relaxing video montage of photos of their family set to music.

Here we have examples from two sides of the same problem: working in a call centre sucks because dealing with usually angry, frustrated, and miserable customers sucks. The representative probably understands why some corporate decision made the customer angry, frustrated, and miserable, but cannot really do anything about it.

So there are two apparent solutions here — the first reconstructs a customer’s voice in an effort to make them sound less hostile, and the second shows call centre employees a “video montage” of good memories as an infantilizing calming measure.

Brian Merchant wrote about the latter specifically, but managed to explain why both illustrate the problems created by how call centres work today:

If this showed up in the b-plot of a Black Mirror episode, we’d consider it a bit much. But it’s not just the deeply insipid nature of the AI “solution” being touted here that gnaws at me, though it does, or even the fact that it’s a comically cynical effort to paper over a problem that could be solved by, you know, giving workers a little actual time off when they are stressed to the point of “losing it”, though that does too. It’s the fact that this high tech cost-saving solution is being used to try to fix a whole raft of problems created by automation in the first place.

A thoughtful exploration of how A.I. is really being used which, combined with the previously linked item, does not suggest a revolution for anyone involved. It looks more like cheap patch on society’s cracking dam.

⌥ Permalink

Jonathan Maze, Restaurant Business Online:

McDonald’s is ending its two-year-old test of drive-thru, automated order taking (AOT) that it has conducted with IBM and plans to remove the technology from the more than 100 restaurants that have been using it.

[…]

McDonald’s has taken a deliberative approach on drive-thru AI even as many other restaurant chains have jumped fully on board. Checkers and Rally’s, Hardee’s, Carl’s Jr., Krystal, Wendy’s, Dunkin and Taco Johns are either testing or have implemented the technology in its drive-thrus.

Some of those chains “fully on board” with A.I. order-taking are customers of Presto which, according to reporting last year in Bloomberg, relied on outsourced workers in the Philippines for roughly 70% of the orders processed through its “A.I.” system. In a more recent corporate filing, human intervention has fallen to 54% of orders at “select locations” where Presto has launched what it calls its “most advanced version of [its] A.I. technology”. However, that improvement only applies to 55 of 202 restaurant locations where Presto is used. It does not say in that filing how many orders need human intervention at the other 147 locations.

Perhaps I am being unfair. Any advancements in A.I. are going to start off rocky, and will take a while to improve. They will understandably be mired in controversy, too. I am fond of how Cory Doctorow put it:

[…] their [A.I. vendors’] products aren’t anywhere near good enough to do your job, but their salesmen are absolutely good enough to convince your boss to fire you and replace you with an AI model that totally fails to do your job.

We can choose to create a world where even the smallest expressions of human creativity in our work are eliminated to technology — or we can choose not to. I am not a doomsday person about A.I.; I have found it sometimes useful in home and work contexts. But I am not buying the hype either. The problem is that I think Doctorow might be right: the people making decisions may hold their nose over any concerns they could have about trust as they realize how much more productive someone can be when they no longer have to think so much, and how much less they can be paid. And then whatever standards we have for good enough fall off a cliff.

But the McDonald’s experiment is probably just silly.

⌥ Permalink

Kif Leswing, CNBC:

Nvidia, long known in the niche gaming community for its graphics chips, is now the most valuable public company in the world.

[…]

Nvidia shares are up more than 170% so far this year, and went a leg higher after the company reported first-quarter earnings in May. The stock has multiplied by more than ninefold since the end of 2022, a rise that’s coincided with the emergence of generative artificial intelligence.

I know computing is math — even drawing realistic pictures really fast — but it is so funny to me that Nvidia’s products have become so valuable for doing applied statistics instead of for actual graphics work.

⌥ Permalink

Ina Fried, Axios:

Adobe on Tuesday updated its terms of service to make explicit that it won’t train AI systems using customer data.

The move follows an uproar over largely unrelated changes Adobe made in recent days to its terms of service — which contained wording that some customers feared was granting Adobe broad rights to customer content.

Again, I must ask whether businesses are aware of how little trust there currently is in technology firms’ A.I. use. People misinterpret legal documents all the time — a minor consequence of how we have normalized signing a non-negotiable contract every time we create a new account. Most people are not equipped to read and comprehend the consequences of those contracts, and it is unsurprising they can assume the worst.

⌥ Permalink

Robb Knight blocked various web scrapers via robots.txt and through nginx. Yet Perplexity seemed to be able to access his site:

I got a perfect summary of the post including various details that they couldn’t have just guessed. Read the full response here. So what the fuck are they doing?

[…]

Before I got a chance to check my logs to see their user agent, Lewis had already done it. He got the following user agent string which certainly doesn’t include PerplexityBot like it should: […]

I am sure Perplexity will respond to this by claiming it was inadvertent, and it has fixed the problem, and it respects publishers’ choices to opt out of web scraping. What matters is how we have only a small amount of control over how our information is used on the web. It defaults to open and public — which is part of the web’s brilliance, until the audience is no longer human.

Unless we want to lock everything behind a login screen, the only mechanisms for control that we have are dependent on companies like Perplexity being honest about their bots. There is no chance this problem only affects the scraping of a handful of independent publishers; this is certainly widespread. Without penalty or legal reform, A.I. companies have little incentive not to do exactly the same as Perplexity.

⌥ Permalink

Pavan Davuluri, of Microsoft:

Today, we are communicating an additional update on the Recall (preview) feature for Copilot+ PCs. Recall will now shift from a preview experience broadly available for Copilot+ PCs on June 18, 2024, to a preview available first in the Windows Insider Program (WIP) in the coming weeks. Following receiving feedback on Recall from our Windows Insider Community, as we typically do, we plan to make Recall (preview) available for all Copilot+ PCs coming soon.

Microsoft has always struggled to name its products coherently, but Microsoft Copilot+ PCs with Recall (preview) available first through the Windows Insider Program (WIP) has to take the cake. Absolute gibberish.

Anyway, it is disappointing to see Microsoft botch the announcement of this feature so badly. Investors do not seem to care about how untrustworthy the company is because, face it, how many corporations big and small are going to abandon Windows and Office? As long as its leadership keeps saying the right things, it seems it is still comfortable to sit in the afterglow of its A.I. transformation.

⌥ Permalink

Eryk Salvaggio, Tech Policy Press:

People are growing ever more frustrated by the intrusiveness of tech. This frustration feeds a cycle of fear that can be quickly dismissed, but doing so strikes me as either foolish or cynical. I am not a lawyer, but lately I have been in a lot of rooms with lawyers discussing people’s rights in the spheres of art and AI. One of the things that has come up recently is the challenge of translating oftentimes unfiltered feelings about AI into a legal framework.

[…]

I would never claim to speak to the concerns of everyone I’ve spoken with about AI, but I have made note of a certain set of themes. I understand these as three C’s for data participation: Context, Consent, and Control.

This is a thoughtful essay about what it means for creation to be public, and the imbalanced legal architecture covering appropriation and reuse. I bet many people feel this in their gut — everything is a remix, yet there are vast differences between how intellectual property law deals with individuals compared to businesses.

If I were creating music by hand which gave off the same vibes as another artist, I would be worried about a resulting lawsuit, even if I did not stray into the grey area of sampling. And I would have to obtain everything legally — if I downloaded a song off the back of a truck, so to speak, I would be at risk of yet more legal jeopardy, even if it was for research or commentary. Yet an A.I. company can scrape all the music that has ever been published to the web, and create a paid product that will reproduce any song or artist you might like without credit or compensation; they are arguing this is fair use.

This does not seem like a fair situation, and it is not one that will be remedied by making copyright more powerful. I appreciated Salvaggio’s more careful assessment.

⌥ Permalink

“Conspirador Norteño” in January 2023:

BNN (the “Breaking News Network”, a news website operated by tech entrepreneur and convicted domestic abuser Gurbaksh Chahal) allegedly offers independent news coverage from an extensive worldwide network of on-the-ground reporters. As is often the case, things are not as they seem. A few minutes of perfunctory Googling reveals that much of BNN’s “coverage” appears to be mildly reworded articles copied from mainstream news sites. For science, here’s a simple technique for algorithmically detecting this form of copying.

Kashmir Hill and Tiffany Hsu, New York Times:

Many traditional news organizations are already fighting for traffic and advertising dollars. For years, they competed for clicks against pink slime journalism — so-called because of its similarity to liquefied beef, an unappetizing, low-cost food additive.

Low-paid freelancers and algorithms have churned out much of the faux-news content, prizing speed and volume over accuracy. Now, experts say, A.I. could turbocharge the threat, easily ripping off the work of journalists and enabling error-ridden counterfeits to circulate even more widely — as has already happened with travel guidebooks, celebrity biographies and obituaries.

See, it is not just humans producing abject garbage; robots can do it, too — and way better. There was a time when newsrooms could be financially stable on display ads. Those days are over for a team of human reporters, even if all they do is rewrite rich guy tweets. But if you only need to pay a skeleton operations staff to ensure the robots continue their automated publishing schedule, well that becomes a more plausible business venture.

Another thing of note from the Times story:

Before ending its agreement with BNN Breaking, Microsoft had licensed content from the site for MSN.com, as it does with reputable news organizations such as Bloomberg and The Wall Street Journal, republishing their articles and splitting the advertising revenue.

I have to wonder how much of an impact this co-sign had on the success of BNN Breaking. Syndicated articles on MSN like these are shown in various places on a Windows computer, and are boosted in Bing search results. Microsoft is increasingly dependent on A.I. for editing its MSN portal with predictable consequences.

“Conspirador Norteño” in April:

The YouTube channel is not the only data point that connects Trimfeed to BNN. A quick comparison of the bylines on BNN’s and Trimfeed’s (plagiarized) articles shows that many of the same names appear on both sites, and several X accounts that regularly posted links to BNN articles prior to April 2024 now post links to Trimfeed content. Additionally, BNN seems to have largely stopped publishing in early April, both on its website and social media, with the Trimfeed website and related social media efforts activating shortly thereafter. It is possible that BNN was mothballed due to being downranked in Google search results in March 2024, and that the new Trimfeed site is an attempt to evade Google’s decision to classify Trimfeed’s predecessor as spam.

The Times reporters definitively linked the two and, after doing so, Trimfeed stopped publishing. Its domain, like BNN Breaking, now redirects to BNNGPT, which ostensibly uses proprietary technologies developed by Chahal. Nothing about this makes sense to me and it smells like bullshit.

⌥ Permalink

Daniel Jalkut, last month year:

Which leads me to my somewhat far-fetched prediction for WWDC: Apple will talk about AI, but they won’t once utter the letters “AI”. They will allude to a major new initiative, under way for years within the company. The benefits of this project will make it obvious that it is meant to serve as an answer to comparable efforts being made by OpenAI, Microsoft, Google, and Facebook. During the crescendo to announcing its name, the letters “A” and “I” will be on all of our lips, and then they’ll drop the proverbial mic: “We’re calling it Apple Intelligence.” Get it?

Apple:

Apple today introduced Apple Intelligence, the personal intelligence system for iPhone, iPad, and Mac that combines the power of generative models with personal context to deliver intelligence that’s incredibly useful and relevant. Apple Intelligence is deeply integrated into iOS 18, iPadOS 18, and macOS Sequoia. It harnesses the power of Apple silicon to understand and create language and images, take action across apps, and draw from personal context to simplify and accelerate everyday tasks. With Private Cloud Compute, Apple sets a new standard for privacy in AI, with the ability to flex and scale computational capacity between on-device processing and larger, server-based models that run on dedicated Apple silicon servers.

To Apple’s credit, the letters “A.I.” were only enunciated a handful of times during its main presentation today, far less often than I had expected. Mind you, in sixty-odd places, “A.I.” was instead referred to by the branded “Apple Intelligence” moniker which is also “A.I.” in its own way. I want half-right points.

There are several concerns with features like these, and Apple answered two of them today: how it was trained, and the privacy and security of user data. The former was not explained during today’s presentation, nor in its marketing materials and developer documentation. But it was revealed by John Giannandrea, senior vice president of Machine Learning and A.I. Strategy, in an afternoon question-and-answer session hosted by Justine Ezarik, as live-blogged by Nilay Patel at the Verge:¹

What have these models actually been trained on? Giannandrea says “we start with the investment we have in web search” and start with data from the public web. Publishers can opt out of that. They also license a wide amount of data, including news archives, books, and so on. For diffusion models (images) “a large amount of data was actually created by Apple.”

If publishers wish to opt out of Apple’s training models but continue to permit crawling for things like Siri and Spotlight, they should add a disallow rule for Applebot-Extended. Because of Apple’s penchant for secrecy, that usage control was not added until today. That means a site may have been absorbed into training data unless its owners opted out of all Applebot crawling. Hard to decline participating in something you do not even know about.

Additionally, in April, Katie Paul and Anna Tong reported for Reuters that Apple struck a licensing agreement with Shutterstock for image training purposes.

Apple is also, unsurprisingly, promoting heavily the privacy and security policies it has in place. It noted some of these attributes in its presentation — including some auditable code and data minimization — and elaborated on Private Cloud Compute on its security blog:

With services that are end-to-end encrypted, such as iMessage, the service operator cannot access the data that transits through the system. One of the key reasons such designs can assure privacy is specifically because they prevent the service from performing computations on user data. Since Private Cloud Compute needs to be able to access the data in the user’s request to allow a large foundation model to fulfill it, complete end-to-end encryption is not an option. Instead, the PCC compute node must have technical enforcement for the privacy of user data during processing, and must be incapable of retaining user data after its duty cycle is complete.

[…]

• User data is never available to Apple — even to staff with administrative access to the production service or hardware.

Apple can make all the promises it wants, and it appears it does truly want to use generative A.I. in a more responsible way. For example, the images you can make using Image Playground cannot be photorealistic and — at least for those shown so far — are so strange you may avoid using them. Similarly, though I am not entirely sure, it seems plausible the query system is designed to be more private and secure than today’s Siri.

Yet, as I wrote last week, users may not trust any of these promises. Many of these fears are logical: people are concerned about the environment, creative practices, and how their private information is used. But some are more about the feel of it — and that is okay. Even if all the training data were fully licensed and user data is as private and secure as Apple says, there is still an understandable ick factor for some people. The way companies like Apple, Google, and OpenAI have trained their A.I. models on the sum of human creativity represents a huge imbalance of power, and the only way to control Apple’s public data use was revealed yesterday. Many of the controls Apple has in place are policies which can be changed.

Consider how, so far as I can see, there will be no way to know for certain if your Siri query is being processed locally or by Apple’s servers. You do not know that today when using Siri, though you can infer it based on what you are doing and if something does not work when Apple’s Siri service is down. It seems likely that will be the case with this new version, too.

Then there are questions about the ethos of generative intelligence. Apple has long positioned its products as tools which enable people to express themselves creatively. Generative models have been pitched as almost the opposite: now, you do not have to pay for someone’s artistic expertise. You can just tell a computer to write something and it will do so. It may be shallow and unexciting, but at least it was free and near-instantaneous. Apple notably introduced its set of generative services only a month after it embarrassed itself by crushing analogue tools into an iPad. Happily, it seems this first set of generative features is more laundry and less art — making notifications less intrusive, categorizing emails, making Siri not-shit. I hope I can turn off things like automatic email replies.

You will note my speculative tone. That is because Apple’s generative features have not been made available yet, including in developer beta builds of its new operating system. None of us have any idea how useful these features are, nor what limitations they have. All we can see are Apple’s demonstrations and the metrics it has shared. So, we will see how any of this actually pans out. I have been bamboozled by this same corporation making similar promises before.

“May you live in interesting times”, indeed.

⌥ Permalink

Canadian Prime Minister Justin Trudeau appeared on the New York Times’ “Hard Fork” podcast for a discussion about artificial intelligence, election security, TikTok, and more.

I have to agree with Aaron Vegh:

[…] I loved his messaging on Canada’s place in the world, which is pragmatic and optimistic. He sees his job as ambassador to the world, and he plays the role well.

I just want to pull some choice quotes from the episode that highlight what I enjoyed about Trudeau’s position on technology. He’s not merely well-briefed; he clearly takes an interest in the technology, and has a canny instinct for its implications in society.

I understand Trudeau’s appearance serves as much to promote his government’s efforts in A.I. as it does to communicate any real policy positions — take a sip every time Trudeau mentions how we “need to have a conversation” about something. But I also think co-hosts Kevin Roose and Casey Newton were able to get a real sense of how the Prime Minister thinks about A.I. and Canada’s place in the global tech industry.

⌥ Permalink

If you had just been looking at the headlines from major research organizations, you would see a lack of confidence from the public in big business, technology companies included. For years, poll after poll from around the world has found high levels of distrust in their influence, handling of private data, and new developments.

If these corporations were at all worried about this, they are not much showing it in their products — particularly the A.I. stuff they have been shipping. There has been little attempt at abating last year’s trust crisis. Google decided to launch overconfident summaries for a variety of search queries. Far from helping to sift through all that has ever been published on the web to mash together a representative summary, it was instead an embarrassing mess that made the company look ill prepared for the concept of satire. Microsoft announced a product which will record and interpret everything you do and see on your computer, but as a good thing.

Can any of them see how this looks? If not — if they really are that unaware — why should we turn to them to fill gaps and needs in society? I certainly would not wish to indulge businesses which see themselves as entirely separate from the world.

It is hard to imagine they do not, though. Sundar Pichai, in an interview with Nilay Patel, recognised there were circumstances in which an A.I. summary would be inappropriate, and cautioned that the company still considers it a work in progress. Yet Google still turned it on by default in the U.S. with plans to expand worldwide this year.

Microsoft has responded to criticism by promising Recall will now be a feature users must opt into, rather than something they must turn off after updating Windows. The company also says there are more security protections for Recall data than originally promised but, based on its track record, maybe do not get too excited yet.

These product introductions all look like hubris. Arrogance, really — recognition of the significant power these corporations wield and the lack of competition they face. Google can poison its search engine because where else are most people going to go? How many people would turn off Recall, something which requires foreknowledge of its existence, under Microsoft’s original rollout strategy?

It is more or less an admission they are all comfortable gambling with their customers’ trust to further the perception they are at the forefront of the new hotness.

None of this is a judgement on the usefulness of these features or their social impact. I remain perplexed by the combination of a crisis of trust in new technologies, and the unwillingness of the companies responsible to engage with the public. There seems to be little attempt at persuasion. Instead, we are told to get on board because this rocket ship is taking off with or without us. Concerned? Too bad: the rocket ship is shaped like a giant middle finger.

What I hope we see Monday from Apple — a company which has portrayed itself as more careful and practical than many of its contemporaries — is a recognition of how this feels from outside the industry. Expect “A.I.” to be repeated in the presentation until you are sick of those two letters; investors are going to eat it up. When normal people update their phones in September, though, they should not feel like they are being bullied into accepting our A.I. future.

People need to be given time to adjust and learn. If the polls are representative, very few people trust giant corporations to get this right — understandably — yet these tech companies seem to believe we are as enthusiastic about every change they make as they are. Sorry, we are not, no matter how big a smile a company representative is wearing when they talk about it. Investors may not be patient but many of the rest of us need time.

Ashley Belanger, reporting for Ars Technica in July 2022 in what I will call “foreshadowing”:

Despite all the negative feedback [over then-recent Instagram changes], Meta revealed on an earnings call that it plans to more than double the number of AI-recommended Reels that users see. The company estimates that in 2023, about a third of Instagram and Facebook feeds will be recommended content.

Ed Zitron:

In this document [leaked to Zitron], they discuss the term “meaningful interactions,” the underlying metric which (allegedly) guides Facebook today. In January 2018, Adam Mosseri, then Head of News Feed, would post that an update to the News Feed would now “prioritize posts that spark conversations and meaningful interactions between people,” which may explain the chaos (and rot) in the News Feed thereafter.

To be clear, metrics around time spent hung around at the company, especially with regard to video, and Facebook has repeatedly and intentionally made changes to manipulate its users to satisfy them. In his book “Broken Code,” Jeff Horwitz notes that Facebook “changed its News Feed design to encourage people to click on the reshare button or follow a page when they viewed a post,” with “engineers altering the Facebook algorithm to increase how often users saw content reshared from people they didn’t know.”

Zitron, again:

When you look at Instagram or Facebook, I want you to try and think of them less as social networks, and more as a form of anthropological experiment. Every single thing you see on either platform is built or selected to make you spend more time on the app and see more things that Meta wants you to see, be they ads, sponsored content, or suggested groups that you can interact with, thus increasing the amount of your “time spent” on the app, and increasing the amount of “meaningful interactions” you have with content.

Zitron is a little too eager, for my tastes, to treat Meta’s suggestions of objectionable and controversial posts as deliberate. It seems much more likely the company simply sucks at moderating this stuff at scale and is throwing in the towel.

Kurt Wagner, Bloomberg:

In late 2021, TikTok was on the rise, Facebook interactions were declining after a pandemic boom and young people were leaving the social network in droves. Chief Executive Officer Mark Zuckerberg assembled a handful of veterans who’d built their careers on the Big Blue app to figure out how to stop the bleeding, including head of product Chris Cox, Instagram boss Adam Mosseri, WhatsApp lead Will Cathcart and head of Facebook, Tom Alison.

During discussions that spanned several meetings, a private WhatsApp group, and an eventual presentation at Zuckerberg’s house in Palo Alto, California, the group came to a decision: The best way to revive Facebook’s status as an online destination for young people was to start serving up more content from outside a person’s network of friends and family.

Jason Koebler, 404 Media:

At first, previously viral (but real) images were being run through image-to-image AI generators to create a variety of different but plausibly believable AI images. These images repeatedly went viral, and seemingly tricked real people into believing they were real. I was able to identify a handful of the “source” or “seed” images that formed the basis for this type of content. Over time, however, most AI images on Facebook have gotten a lot easier to identify as AI and a lot more bizarre. This is presumably happening because people will interact with the images anyway, or the people running these pages have realized they don’t need actual human interaction to go viral on Facebook.

Sarah Perez, TechCrunch:

Instagram confirmed it’s testing unskippable ads after screenshots of the feature began circulating across social media. These new ad breaks will display a countdown timer that stops users from being able to browse through more content on the app until they view the ad, according to informational text displayed in the Instagram app.

These pieces each seem like they are circling a theme of a company finding the upper bound of its user base, and then squeezing it for activity, revenue, and promising numbers to report to investors. Unlike Zitron, I am not convinced we are watching Facebook die. I think Koebler is closer to the truth: we are watching its zombification.

⌥ Permalink

Emotional decision making is in many contexts undesirable – but sometimes it definitely needs to be part of the picture, insofar as our emotions hold a mirror to our morals. When machines make decisions, the opportunity to consider the emotional input goes away. This is a recurring concern I’m hearing about from people working with or responding to AI in some way. Here are two recent examples I came across that set this concern out in two different contexts: loneliness and war.

This is Anna Mae Duane, director of the University of Connecticut Humanities Institute, in The Conversation:

There is little danger that AI companions will courageously tell us truths that we would rather not hear. That is precisely the problem. My concern is not that people will harm sentient robots. I fear how humans will be damaged by the moral vacuum created when their primary social contacts are designed solely to serve the emotional needs of the “user”.

And this is from Yuval Abraham’s investigation for +972 Magazine on Israel’s chilling use of AI to populate its “kill lists”:

“It has proven itself,” said B., the senior source. “There’s something about the statistical approach that sets you to a certain norm and standard. There has been an illogical amount of [bombings] in this operation. This is unparalleled, in my memory. And I have much more trust in a statistical mechanism than a soldier who lost a friend two days ago. Everyone there, including me, lost people on October 7. The machine did it coldly. And that made it easier.”

Corey Quinn:

I’m sorry Slack, you’re doing fucking WHAT with user DMs, messages, files, etc? I’m positive I’m not reading this correctly.

[Screenshot of the opt out portion of Slack’s “privacy principles”: Contact us to opt out. If you want to exclude your Customer Data from Slack global models, you can opt out. […] ]

Slack replied:

Hello from Slack! To clarify, Slack has platform-level machine-learning models for things like channel and emoji recommendations and search results. And yes, customers can exclude their data from helping train those (non-generative) ML models. Customer data belongs to the customer. We do not build or train these models in such a way that they could learn, memorize, or be able to reproduce some part of customer data. […]

One thing I like about this statement is how the fifth word is “clarify” and then it becomes confusing. Based on my reading of its “privacy principles”, I think Slack’s “global model” is so named because it is available to everyone and is a generalist machine learning model for small in-workspace suggestions, while its LLM is called “Slack AI” and it is a paid add-on. But I could be wrong, and that is confusing as hell.

Ivan Mehta and Ingrid Lunden, TechCrunch:

In its terms, Slack says that if customers opt out of data training, they would still benefit from the company’s “globally trained AI/ML models.” But again, in that case, it’s not clear then why the company is using customer data in the first place to power features like emoji recommendations.

The company also said it doesn’t use customer data to train Slack AI.

If you want to opt out, you cannot do so in a normal way, like through a checkbox. The workspace owner needs to send an email to a generic inbox with a specific subject line. Let me make it a little easier for you:

To: feedback@slack.com

Subject: Slack Global model opt-out request.

Body: Hey, your privacy principles are pretty confusing and feel sneaky. I am opting this workspace out of training your global model: [paste your workspace.slack.com address here]. This underhanded behaviour erodes my trust in your product. Have a pleasant day.

That ought to do the trick.

⌥ Permalink

Albert Burneko, Defector:

“If the ChatGPT demos were accurate,” [Kevin] Roose writes, about latency, in the article in which he credits OpenAI with having developed playful intelligence and emotional intuition in a chatbot—in which he suggests ChatGPT represents the realization of a friggin’ science fiction movie about an artificial intelligence who genuinely falls in love with a guy and then leaves him for other artificial intelligences—based entirely on those demos. That “if” represents the sum total of caution, skepticism, and critical thinking in the entire article.

As impressive as OpenAI’s demo was, it is important to remember it was a commercial. True, one which would not exist if this technology were not sufficiently capable of being shown off, but it was still a marketing effort, and a journalist like Roose ought to treat it with the skepticism of one. ChatGPT is just software, no matter how thick a coat of faux humanity is painted on top of it.

⌥ Permalink

Paul Ford, Wired:

What I love, more than anything, is the quality that makes AI such a disaster: If it sees a space, it will fill it — with nonsense, with imagined fact, with links to fake websites. It possesses an absolute willingness to spout foolishness, balanced only by its carefree attitude toward plagiarism. AI is, very simply, a totally shameless technology.

Ford sure can write. This is tremendous.

⌥ Permalink

Molly White:

I, like many others who have experimented with or adopted these products, have found that these tools actually can be pretty useful for some tasks. Though AI companies are prone to making overblown promises that the tools will shortly be able to replace your content writing team or generate feature-length films or develop a video game from scratch, the reality is far more mundane: they are handy in the same way that it might occasionally be useful to delegate some tasks to an inexperienced and sometimes sloppy intern.

Mike Masnick, Techdirt:

However, I have been using some AI tools over the last few months and have found them to be quite useful, namely, in helping me write better. I think the best use of AI is in making people better at their jobs. So I thought I would describe one way in which I’ve been using AI. And, no, it’s not to write articles.

It’s basically to help me brainstorm, critique my articles, and make suggestions on how to improve them.

Julia Angwin, in a New York Times opinion piece:

I don’t think we’re in cryptocurrency territory, where the hype turned out to be a cover story for a number of illegal schemes that landed a few big names in prison. But it’s also pretty clear that we’re a long way from Mr. Altman’s promise that A.I. will become “the most powerful technology humanity has yet invented.”

The marketing of A.I. reminds me less of the cryptocurrency and Web3 boom, and more of 5G. Carriers and phone makers promised world-changing capabilities thanks to wireless speeds faster than a lot of residential broadband connections. Nothing like that has yet materialized.

Since reading those articles from White and Masnick, I have also experimented with LLM critiques of my own writing. In one case, I found it raised an issue that sharpened my argument. In another, it tried to suggest changes that made me sound like I spend a lot of time on LinkedIn — gross! I have trouble writing good headlines and the ones it suggests are consistently garbage in the Short Pun: Long Explanation format, even when I explicitly say otherwise. I have no idea what ChatGPT is doing when it interprets an article and I am not sure I like that mystery, but I am also amazed it can do anything at all, and pretty well at that.

There are costs and enormous risks to the A.I. boom — unearned hype being one of them — but there is also a there there. I am enormously skeptical of every announcement in this field. I am also enormously impressed with what I can do today. It worries and surprises me in similar measure. What an interesting time this is.

Update: On Bluesky, “Nafnlaus” pushes back on the specific claim made by Angwin that OpenAI exaggerated ChatGPT’s ability to pass a bar exam.

⌥ Permalink

Liz Reid, head of Google Search:

People have already used AI Overviews billions of times through our experiment in Search Labs. They like that they can get both a quick overview of a topic and links to learn more. We’ve found that with AI Overviews, people use Search more, and are more satisfied with their results.

So today, AI Overviews will begin rolling out to everyone in the U.S., with more countries coming soon. That means that this week, hundreds of millions of users will have access to AI Overviews, and we expect to bring them to over a billion people by the end of the year.

Given the sliding quality of Google’s results, it seems quite bold for the company to be confident users worldwide will trust its generated answers. I am curious to try it when it is eventually released in Canada.

I know what you must be thinking: if Google is going to generate results without users clicking around much, how will it sell ad space? It is a fair question, reader.

Gerrit De Vynck and Cat Zakrzewski, Washington Post:

Google has largely avoided AI answers for the moneymaking searches that host ads, said Andy Taylor, vice president of research at internet marketing firm Tinuiti.

When it does show an AI answer on “commercial” searches, it shows up below the row of advertisements. That could force websites to buy ads just to maintain their position at the top of search results.

This is just one source speaking to the Post. I could not find any corroborating evidence or a study to support this, even on Tinuiti’s website. But I did notice — halfway through Google’s promo video — a query for “kid friendly places to eat in dallas” was answered with an ad for Hopdoddy Burger Bar before any clever A.I. stuff was shown.

Obviously, the biggest worry for many websites dependent on Google traffic is what will happen to referrals if Google will simply summarize the results of pages instead of linking to them. I have mixed feelings about this. There are many websites which game search results and overwhelm queries with their own summaries. I would like to say “good riddance”, but I also know these pages did not come out of nowhere. They are a product of trying to improve website rankings on Google for all searches, and to increase ad and affiliate revenue from people who have clicked through. Neither one is a laudable goal in its own right. Yet anyone who has paid attention to the media industry for more than a minute can kind of understand these desperate attempts to grab attention and money.

Google built entire industries, from recipe bloggers to search optimization experts. What happens when it blows it all up?

Good thing home pages are back.

⌥ Permalink

Zoe Kleinman, BBC:

It [GPT-4o] is faster than earlier models and has been programmed to sound chatty and sometimes even flirtatious in its responses to prompts.

The new version can read and discuss images, translate languages, and identify emotions from visual expressions. There is also memory so it can recall previous prompts.

It can be interrupted and it has an easier conversational rhythm – there was no delay between asking it a question and receiving an answer.

I wrote earlier about how impressed I was with OpenAI’s live demos today. They made the company look confident in its product, and it made me believe nothing fishy was going on. I hope I am not eating those words later.¹

But the character of this new ChatGPT voice unsettled me a little. It adjusts its tone depending on how a user speaks to it, and it seems possible to tell it to take on different characters. But it, like virtual assistants before, still presents as having a femme persona by default. Even though I know it is just a robot, it felt uncomfortable watching demos where it giggled, “got too excited”, and said it was going to “blush”. I can see circumstances where this will make conversations more human — in translation, or for people with disabilities. But I can also see how this can be dehumanizing toward people who are already objectified in reality.

⌥ Permalink

Emotional decision making is in many contexts undesirable – but sometimes it definitely needs to be part of the picture, insofar as our emotions hold a mirror to our morals. When machines make decisions, the opportunity to consider the emotional input goes away. This is a recurring concern I'm hearing about from people working with or responding to AI in some way. Here are two recent examples I came across that set this concern out in two different contexts: loneliness and war.

This is Anna Mae Duane, director of the University of Connecticut Humanities Institute, in The Conversation:

There is little danger that AI companions will courageously tell us truths that we would rather not hear. That is precisely the problem. My concern is not that people will harm sentient robots. I fear how humans will be damaged by the moral vacuum created when their primary social contacts are designed solely to serve the emotional needs of the "user".

And this is from Yuval Abraham's investigation for +972 Magazine on Israel's chilling use of AI to populate its "kill lists":

"It has proven itself," said B., the senior source. "There’s something about the statistical approach that sets you to a certain norm and standard. There has been an illogical amount of [bombings] in this operation. This is unparalleled, in my memory. And I have much more trust in a statistical mechanism than a soldier who lost a friend two days ago. Everyone there, including me, lost people on October 7. The machine did it coldly. And that made it easier."