March 2023 - net.wars

Unclear and unpresent dangers

Monthly computer magazines used to fret that their news pages would be out of date by the time the new issue reached readers. This week in AI, a blog posting is out of date before you hit send.

This – Friday – morning, the Italian data protection authority, Il Garante, has ordered ChatGPT to stop processing the data of Italian users until it complies with the General Data Protection Regulation. Il Garante’s objections, per Apple’s translation, posted by Ian Brown: ChatGPT provides no legal basis for collecting and processing its massive store of the personal data used to train the model, and that it fails to filter out users under 13.

This may be the best possible answer to the complaint I’d been writing below.

On Wednesday, the Future of Life Institute published an open letter calling for a six-month pause on developing systems more powerful than Open AI’s current state of the art, GPT4. Barring Elon Musk, Steve Wozniack, and Skype co-founder Jaan Tallinn, most of the signatories are unfamiliar names to most of us, though the companies and institutions they represent aren’t – Pinterest, the MIT Center for Artificial Intelligence, UC Santa Cruz, Ripple, ABN-Amro Bank. Almost immediately, there was a dispute over the validity of the signatures..

My first reaction was on the order of: huh? The signatories are largely people who are inventing this stuff. They don’t have to issue a call. They can just *stop*, work to constrain the negative impacts of the services they provide, and lead by example. Or isn’t that sufficiently performative?

A second reaction: what about all those AI ethics teams that Silicon Valley companies are disbanding? Just in the last few weeks, these teams have been axed or cut at Microsoft and Twitch; Twitter of course ditched such fripperies last November in Musk’s inaugural wave of cost-cutting. The letter does not call to reinstate these.

The problem, as familiar critics such as Emily Bender pointed out almost immediately, is that the threats the letter focuses on are distant not-even-thunder. As she went on to say in a Twitter thread, the artificial general intelligence of the Singularitarian’s rapture is nowhere in sight. By focusing on distant threats – longtermism – we ignore the real and present problems whose roots are being continuously more deeply embedded into the new-building infrastructure: exploited workers, culturally appropriated data, lack of transparency around the models and algorithms used to build these systems….basically, all the ways they impinge upon human rights.

This isn’t the first time such a letter has been written and circulated. In 2015, Stephen Hawking, Musk, and about 150 others similarly warned of the dangers of the rise of “superintelligences”. Just a year later, in 2016, Pro Publica investigated the algorithm behind COMPAS, a risk-scoring criminal justice system in use in US courts in several states. Under Julia Angwin‘s scrutiny, the algorithm failed at both accuracy and fairness; it was heavily racially biased. *That*, not some distant fantasy, was the real threat to society.

“Threat” is the key issue here. This is, at heart, a letter about a security issue, and solutions to security issues are – or should be – responses to threat models. What is *this* threat model, and what level of resources to counter it does it justify?

Today, I’m far more worried by the release onto public roads of Teslas running Full Self Drive helmed by drivers with an inflated sense of the technology’s reliability than I am about all of human work being wiped away any time soon. This matters because, as Jessie Singal, author of There Are No Accidents, keeps reminding us, what we call “accidents” are the results of policy decisions. If we ignore the problems we are presently building in favor of fretting about a projected fantasy future, that, too, is a policy decision, and the collateral damage is not an accident. Can’t we do both? I imagine people saying. Yes. But only if we *do* both.

In a talk this week for a group at the French international research group AI Act. This effort began well before today’s generative tools exploded into public consciousness, and isn’t likely to conclude before 2024. It is, therefore, much more focused on the kinds of risks attached to public sector scandals like COMPAS and those documented in Cathy O’Neil’s 2017 book Weapons of Math Destruction, which laid bare the problems with algorithmic scoring with little to tether it to reality.

With or without a moratorium, what will “AI” look like in 2024? It has changed out of recognition just since the last draft text was published. Prediction from this biological supremacist: it still won’t be sentient.

All this said, as Edwards noted, even if the letter’s proposal is self-serving, a moratorium on development is not necessarily a bad idea. It’s just that if the risk is long-term and existential, what will six months do? If the real risk is the hidden continued centralization of data and power, then those six months could be genuinely destructive. So far, it seems like its major function is as a distraction. Resist.

Illustrations: IBM’s Watson, which beat two of Jeopardy‘s greatest champions in 2011. It has since failed to transform health care.

Wendy M. Grossman is the 2013 winner of the Enigma Award. Her Web site has an extensive archive of her books, articles, and music, and an archive of earlier columns in this series. Follow on Mastodon or Twitter.

Memex 2.0

As language models get cheaper, it’s dawned on me what kind of “AI” I’d like to have: a fully personalized chat bot that has been trained on my 30-plus years of output plus all the material I’ve read, watched, listened to, and taken notes on all these years. A clone of my brain, basically, with more complete and accurate memory updated alongside my own. Then I could discuss with it: what’s interesting to write about for this week’s net.wars?

I was thinking of what’s happened with voice synthesis. In 2011, it took the Scottish company Cereproc months to build a text-to-speech synthesizer from recordings of Roger Ebert’s voice. Today, voice synthesizers are all over the place – not personalized like Ebert’s, but able to read a set text plausibly enough to scare voice actors.

I was also thinking of the Stochastic Parrots paper, whose first anniversary was celebrated last week by authors Emily Bender, Timnit Gebru, Angelina McMillan-Major, and Margaret Mitchell. An important part of the paper advocates for smaller, better-curated language models: more is not always better. I can’t find a stream for the event, but here’s the reading list collected during the proceedings. There’s lots I’d rather eliminate from my personal assistant. Eliminating unwanted options upfront has long been a widspread Internet failure, from shopping sites (“never show me pet items”) to news sites (“never show me fashion trends”). But that sort of selective display is more difficult and expensive than including everything and offering only inclusion filters.

A computational linguistics expert tells me that we’re an unknown amount of time away from my dream of the wg-bot. Probably, if such a thing becomes possible it will be based on someone’s large language model and fine-tuned with my stuff. Not sure I entirely like this idea; it means the model will be trained on stuff I haven’t chosen or vetted and whose source material is unknown, unless we get a grip on forcing disclosure or the proposed BLOOM academic open source language model takes over the world.

I want to say that one advantage to training a chatbot on your own output is you don’t have to worry so much about copyright. However, the reality is that most working writers have sold all rights to most of their work to large publishers, which means that such a system is a new version of digital cholera. In my own case, by the time I’d been in this business for 15 years, more than half of the publications I’d written for were defunct. I was lucky enough to retain at least non-exclusive rights to my most interesting work, but after so many closures and sales I couldn’t begin to guess – or even know how to find out – who owns the rights to the rest of it. The question is moot in any case: unless I choose to put those group reviews of Lotus 1-2-3 books back online, probably no one else will, and if I do no one will care.

On Mastodon, the specter of the upcoming new! improved! version of the copyright wars launched by the arrival of the Internet: “The real generative AI copyright wars aren’t going to be these tiny skirmishes over artists and Stability AI. Its going to be a war that puts filesharing 2.0 and the link tax rolled into one in the shade.” Edwards is referring to this case, in which artists are demanding billions from the company behind the Stable Diffusion engine.

Edwards went on to cite a Wall Street Journal piece that discusses publishers’ alarmed response to what they perceive as new threats to their business. First: that the large piles of data used to train generative “AI” models are appropriated without compensation. This is the steroid-fueled analogue to the link tax, under which search engines in Australia pay newspapers (primarily the Murdoch press) for including them in news search results. A similar proposal is pending in Canada.

The second is that users, satisfied with the answers they receive from these souped-up search services will no longer bother to visit the sources – especially since few, most notably Google, seem inclined to offer citations to back up any of the things they say.

The third is outright plagiarism without credit by the chatbot’s output, which is already happening.

The fourth point of contention is whether the results of generative AI should be themselves subject to copyright. So far, the consensus appears to be no, when it comes to artwork. But some publishers who have begun using generative chatbots to create “content” no doubt claim copyright in the results. It might make more sense to copyright the *prompt*. (And some bright corporate non-soul may yet try.)

At Walled Culture, Glyn Moody discovers that the EU has unexpectedly done something right by requiring positive opt-in to copyright protection against text and data mining. I’d like to see this as a ray of hope for avoiding the worst copyright conflicts, but given the transatlantic rhetoric around privacy laws and data flows, it seems much more likely to incite another trade conflict.

It now dawns on me that the system I outlined in the first paragraph is in fact Vannevar Bush’s Memex. Not the web, which was never sufficiently curated, but this, primed full of personal intellectual history. The “AI” represents those thousands of curating secretaries he thought the future would hold. As if.

Illustrations: Stable Diffusion rendering of “stochastic parrots”, as prompted by Jon Crowcroft.

Performing intelligence

“Oh, great,” I thought when news broke of the release of GPT-4. “Higher-quality deception.”

Most of the Internet disagreed; having gone mad only a few weeks ago over ChatGPT, everyone’s now agog over this latest model. It passed all these tests!

One exception was the journalist Paris Marx, who commented on Twitter: “It’s so funny to me that the AI people think it’s impressive when their programs pass a test after being trained on all the answers.”

Agreed. It’s also so funny to me that they call that “AI” and don’t like it when researchers like computational linguist Emily Bender call it a “stochastic parrot”. At Marx’s Tech Won’t Save Us podcast, Goldsmith professor Dan McQuillan, author of Resisting AI: An Anti-fascist Approach to Artificial Intelligence, calls it a “bullshit engine” whose developers’ sole goal is plausibility – plausibility that, as Bender has said, allows us imaginative humans to think we detect a mind behind it, and the result is to risk devaluing humans.

Let’s walk back to an earlier type of system that has been widely deployed: benefits scoring systems. A couple of weeks ago, Lighthouse Reports and Wired magazine teamed up on an investigation of these systems, calling them “suspicion machines”.

Their work focuses on the welfare benefits system in use in Rotterdam between 2017 and 2021, which uses 315 variables to risk-score benefits recipients according to the likelihood that their claims are fraudulent. In detailed, worked case analyses, they find systemic discrimination: you lose points for being female, for being female and having children (males aren’t asked about children), for being non-white, and for ethnicity (knowing Dutch a requirement for welfare recipients). Other variables include missing meetings, age, and “lacks organizing skills”, which was just one of 54 variables based on case workers’ subjective assessments. Any comment a caseworker adds translates to a 1 added to the risk score, even if it’s positive. The top-scoring 10% are flagged for further investigation.

This is the system that Accenture, the city’s technology partner on the early versions, said at its unveiling in 2018 was an “ethical solution” and promised “unbiased citizen outcomes”. Instead, Wired says, the algorithm “fails the city’s own test of fairness”.

The project’s point wasn’t to pick on Rotterdam; of the dozens of cities they contacted it just happened to be the only one that was willing to share the code behind the algorithm, along with the list of variables, prior evaluations, and the data scientists’ handbook. It even – after being threatened with court action under freedom of information laws, shared the mathematical model itself.

The overall conclusion: the system was so inaccurate it was little better than random sampling “according to some metrics”.

What strikes me, aside from the details of this design, is the initial choice of scoring benefits recipients for risk of fraud. Why not score them for risk of missing out on help they’re entitled to? The UK government’s figures on benefits fraud indicate that in 2021-2022 overpayment (including error as well as fraud) amounted to 4%; and *underpayment* 1.2% of total expenditure. Underpayment is a lot less, but it’s still substantial (£2.6 billion). Yes, I know, the point of the scoring system is to save money, but the point of the *benefits* system is to help people who need it. The suspicion was always there, but the technology has altered the balance.

This was the point the writer Ellen Ullman noted in her 1996 book Close to the Machine”: the hard-edged nature of these systems and their ability to surveil people in new ways, “infect” their owners with suspicion even of people they’ve long trusted and even when the system itself was intended to be helpful. On a societal scale, these “suspicion machines” embed increased division in our infrastructure; in his book, McQuillan warns us to watch for “functionality that contributes to violent separations of ‘us and them’.”

Along those lines, it’s disturbing that Open AI, the owner of ChatGPT and GPT-4 (and several other generative AI gewgaws) has now decided to keep secret the details of its large language models. That is, we have no sight into what data was used in training, what software and hardware methods were used, or how energy-intensive it is. If there’s a machine loose in the world’s computer systems pretending to be human, shouldn’t we understand how it works? It would help with damping down imagining we see a mind in there.

The company’s argument appears to be that because these models could become harmful it’s bad to publish how they work because then bad actors will use them to create harm. In the cybersecurity field we call this “security by obscurity” and there is a general consensus that it does not work as a protection.

In a lengthy article at New York magazine, Elizabeth Weil. quotes Daniel Dennett’s assessment of these machines: “counterfeit people” that should be seen as the same sort of danger to our system as counterfeit money. Bender suggests that rather than trying to make fake people we should be focusing more on making tools to help people.

The thing that makes me tie it to the large language models that are producing GPT is that in both cases it’s all about mining our shared cultural history, with all its flaws and misjudgments, in response to a prompt and pretending the results have meaning and create new knowledge. And *that’s* what’s being embedded into the world’s infrastructure. Have we learned nothing from Clever Hans?

Illustrations: Clever Hans, performing in Leipzig in 1912 (by Karl Krali, via Wikimedia.

Gap year

What do Internet users want?

First, they want meaningful access. They want usability. They want not to be scammed, manipulated, lied to, exploited, or cheated.

It’s unlikely that any of the ongoing debates in either the US or UK will deliver any of those.

First and foremost, this week concluded two frustrating years in which the US Senate failed to confirm the appointment of Public Knowledge co-founder and EFF board member Gigi Sohn to the Federal Communications Commission. In her withdrawal statement, Sohn blamed a smear campaign by “legions of cable and media industry lobbyists, their bought-and-paid-for surrogates, and dark money political groups with bottomless pockets”.

Whether you agree or not, the result remains that for the last two years and for the foreseeable future the FCC will remain deadlocked and problems such as the US’s lack of competition and patchy broadband provision will remain unsolved.

Meanwhile, US politicians continue obsessing about whether and how to abort-retry-fail Section 230, that pesky 26-word law that relieves Internet hosts of liability for third-party content. This week it was the turn of the Senate Judiciary Committee. In its hearing, the Internet Society’s Andrew Sullivan stood out for trying to get across to lawmakers that S230 wasn’t – couldn’t have been – intended as protectionism for the technology giants because they did not exist when the law was passed. It’s fair to say that S230 helped allow the growth of *some* Internet companies – those that host user-generated content. That means all the social media sites as well as web boards and blogs and Google’s search engine and Amazon’s reviews, but neither Apple nor Netflix makes its living that way. Attacking the technology giants is a popular pasttime just now, but throwing out S230 without due attention to the unexpected collateral damage will just make them bigger.

Also on the US political mind is a proposed ban on TikTok. It’s hard to think of a move that would more quickly alienate young people. Plus, it fails to get at the root problem. If the fear is that TikTok gathers data on Americans and sends it home to China for use in designing manipulative programs…well, why single out TikTok when it lives in a forest of US companies doing the same kind of thing? As Karl Bode writes at TechDirt, if you really want to mitigate that threat, rein in the whole forest. Otherwise, if China really wants that data it can buy it on the open market.

Meanwhile, in the UK, as noted last week, opposition continues to increase to the clauses in the Online Safety bill proposing to undermine end-to-end encryption by requiring platforms to proactively scan private messages. This week, WhatsApp said it would withdraw its app from the UK rather than comply. However important the UK market is, it can’t possibly be big enough for Meta to risk fines of 4% of global revenues and criminal sanctions for executives. The really dumb thing is that everyone within the government uses WhatsApp because of its convenience and security, and we all know it. Or do they think they’ll have special access denied the rest of the population?

Also in the UK this week, the Data Protection and Digital Information bill returned to Parliament for its second reading. This is the UK’s post-Brexit attempt to “take control” by revising the EU’s General Data Protection Regulation; it was delayed during Liz Truss’s brief and destructive outing as prime minister. In its statement, the government talks about reducing the burdens on businesses without any apparent recognition that divergence from GDPR is risky for anyone trading internationally and complying with two regimes must inevitably be more expensive than complying with one.

The Open Rights Group and 25 other civil society organizations have written a letter (PDF) laying out their objections, noting that the proposed bill, in line with other recent legislation that weakens civil rights, weakens oversight and corporate accountability, lessens individuals’ rights, and weakens the independence of the Information Commissioner’s Office. “Co-designed with businesses from the start” is how the government describes the bill. But data protection law was not supposed to be designed for business – or, as Peter Geoghegan says at the London Review of Books, to aid SLAPP suits; it is supposed to protect our human rights in the face of state and corporate power. As the cryptography pioneer Whit Diffie said in 2019, “The problem isn’t privacy; it’s corporate malfeasance.”

The most depressing thing about all of these discussions is that the public interest is the loser in all of them. It makes no sense to focus on TikTok when US companies are just as aggressive in exploiting users’ data. It makes no sense to focus solely on the technology giants when the point of S230 was to protect small businesses, non-profits, and hobbyists. And it makes no sense to undermine the security afforded by end-to-end encryption when it’s essential for protecting the vulnerable people the Online Safety bill is supposed to help. In a survey, EDRi finds that compromising secure messaging is highly unpopular with young people, who clearly understand the risks to political activism and gender identity exploration.

One of the most disturbing aspects of our politics in this century so far is the widening gap between what people want, need, and know and the things politicians obsess about. We’re seeing this reflected in Internet policy, and it’s not helpful.

Illustrations: Andrew Sullivan, president of the Internet Society, testifying in front of the Senate Judiciary Committee.

Ghostwritten

This week’s deliberate leak of 100,000 WhatsApp messages sent between the retiring MP Matt Hancock (Con-West Suffolk) and his cabinet colleagues and scientific advisers offers several lessons for the future. Hancock was the health minister during the first year of the covid-19 pandemic, but forced to resign in June 2021, when he was caught on a security camera snogging an adviser in contravention of the social distancing rules.

The most ignored lesson relates to cybersecurity, and is simple: electronic messages are always at risk of copying and disclosure.

This leak happened to coincide with the revival of debates around the future of strong encryption in the UK. First, the pending Online Safety bill has provisions that taken together would undermine all encrypted communications. Simultaneously, a consultation on serious and organized crime proposes to criminalize “custom” encryption devices. A “dictionary attack”, Tim Cushing calls this idea at Techdirt, in that the government will get to define the crime at will.

The Online Safety Bill is the more imminent problem; it has already passed the House of Commons and is at the committee stage in the House of Lords. The bill requires service providers to protect children by proactively removing harmful content, whether public or private, and threatens criminal liability for executives of companies that fail to comply.

Signal, which is basically the same as WhatsApp without the Facebook ownership, has already said it will leave the country if the Online Safety bill passes with the provisions undermining encryption intact.

It’s hard to see what else Signal could do. It’s not a company that has to weigh its principles against the loss of revenue. Instead, as a California non-profit, its biggest asset is the trust of its user base, and staying in a country that has outlawed private communications would kill that off at speed. In threatening to leave it has company: the British secure communications company Element, which said the provisions would taint any secure communications product coming out of the UK – presumably even for its UK customers, such as the Ministry of Defence.

What the Hancock leak reminds us, however, is that encryption, even when appropriately strong and applied end-to-end, is not enough by itself to protect security. You must also be able to trust everyone in the chain to store the messages safely and respect their confidentiality. The biggest threat is careless or malicious insiders, who can undermine security in all sorts of ways. Signal (as an example) provides the ability to encrypt the message database, to disappear messages on an automated schedule, password protection, and so on. If you’re an activist in a hostile area, you may be diligent about turning all these on. But you have no way of knowing if your correspondents are just as careful.

In the case at hand, Hancock gave the messages to the ghost writer for his December 2022 book Pandemic Diaries, Isabel Oakeshott, after requiring her to sign a non-disclosure agreement that he must have thought would protect him, if not his colleagues, from unwanted disclosures. Oakeshott, who claims she acted in the public interest, decided to give the messages to the Daily Telegraph, which is now mining them for stories.

Digression: whatever Oakeshott’s personal motives, there is certainly public interest in these messages. The tone of many quoted exchanges confirms the public perception of the elitism and fecklessness of many of those in government. More interesting is the close-up look at decision making in conditions of uncertainty, which to some filled with hindsight looks like ignorance and impatience. It’s astonishing how quickly people have forgotten how much we didn’t know. As mathematician Christina Pagel told the BBC’s Newsnight, you can’t wait for more evidence when the infection rate is doubling every four days.

What they didn’t know and when they didn’t know it will be an important part of piecing together what actually happened. The mathematician Kit Yates has dissected another exchange, in which Boris Johnson queries his scientific advisers about fatality rates. Yates argues that in assessing this exchange timing ise everything. Had it been in early 2020, it would be understandable to confuse infection fatality rates and case fatality rates, though less so to confuse fractions (0.04) and percentages (4%). Yates pillories Johnson because in fact that exchange took place in August 2020, by which time greater knowledge should have conferred greater clarity. That said, security people might find familiar Johnson’s behavior in this exchange, where he appears to see the Financial Times as a greater authority than the scientists. Isn’t that just like every company CEO?

Exchanges like that are no doubt why the participants wanted the messages kept private. In a crisis, you need to be able to ask stupid questions. It would be better to have a prime minister who can do math and who sweats the details, but if that’s not what we’ve got I’d rather he at least asked for clarification.

Still, as we head into yet another round of the crypto wars, the bottom line is this: neither technology nor law prevented these messages from leaking out some 30 years early. We need the technology. We need the law on our side. But even then, your confidences are only ever as private as your correspondent(s) and their trust network(s) will allow.

Illustrations: The soon-to-be-former-MP Matt Hancock, on I’m a Celebrity.