Tag: Grok

In search of causality

The debates over children’s use of social media, screens, and phones continue, exacerbated in the UK by ongoing Parliamentary scrutiny of the Children’s Wellbeing and Schools bill and continuing disgust over Grok‘s sexualized image generation. Robert Booth reports at the Guardian that the Center for Countering Digital Hate estimates that Grok AI generated 3 million sexualized images in under two weeks and that a third of them are still viewable on X. In that case, X and Grok appear to be a more general problem than children’s access.

We continue to need better evidence establishing causality or its absence. This week, researchers from the Bradford Centre for Health Data Science (led by Dan Lewer) and the University of Cambridge (led by Amy Orben) announced a six-week trial that will attempt to find the actual impact on teens of limiting – not ending – social media access. The BBC reports, that the trial will split 4,000 Bradford secondary school pupils into two groups, One will download an app hat turns off access to services like TikTok and Snapchat from 9pm to 7am and limits use at other times to a “daily budget”. The restrictions won’t include WhatsApp, which the researchers recognize is central to many family groups. The other half will go on using social media as before.

The researchers will compare the two groups by assessing their’ levels of anxiety, depression, sleep, bullying, and time spent with friends and family.

In earlier research, Orben developed a framework for data donation, which allows teens to understand their own use of social media. Another forthcoming study, Youth Perspectives on Social Media Harms: A Large-Scale Micro-Narrative Study, collects 901 first-person tales from 18- to 22-year-olds in the UK. From these Orben’s group derive four types of harm: harms from other people’s behavior, personal harmful behavior evoked by social media, harms related to the content they encounter, and harms related to platform features. In the first category they include bullying and scams; in the second, compulsive use and social comparison; in the third, graphic material; and in the fourth, algorithmic manipulation. They also note the study’s limitations. A longer-term or differently-timed study might show different effects – during the study period the 2024 US presidential election took place. The teens’ stories don’t establish causality. Finally, there may be other harms not captured in this study.

The most important element, however, is that they sought the perspective of young people themselves, who are to date rarely heard in these discussions.

As this research begins, at Techdirt Mike Masnick reports on two new finished papers also covering teens and social media. The first, Social Media Use and Well-Being Across Adolescent Development, published in JAMA Pediatrics, is a three-year study of 100,991 Australian adolescents to find whether well-being was associated with social media use. The researchers, from the University of South Australia, found a U-shaped curve: moderate social media use was associated with the best outcomes, while both the highest users and the non-users showed less well-being. Girls benefited increasingly from moderate social media use from mid-adolescence onwards, while in boys’ non-use became increasingly problematic, leading to worse outcomes than high use by their late teens.

The second, a study from the University of Manchester published in the Journal of Public Health, followed a group of 25,000 11- to 14-year-olds to find out whether the use of technology such as social media and gaming accurately predicted later mental health issues. The study found no evidence that heavier use of social media or gaming led to increased symptoms of anxiety or depression in the following year.

In his discussion of these two papers, Masnick argues that this research gives weight to his contention that the widespread claim that social media is inherently harmful is wrong.

In the UK and elsewhere, however, politicians are proceeding on the basis that social media *is* inevitably harmful. . This week, the government announced a consultation on children’s use of technology. The consultation seems, as Carly Page writes at The Register, geared toward increasing restrictions, Also this week, the House of Lords voted 261 to 150, defeating the government to add an amendment to the Children’s Wellbeing and Schools bill that would require social media services to add age verification to block under-16s from accessing them within a year. MPs will now have to vote to remove the amendment or it will become law, a backdoor preemption of the House of Commons’ prerogative to legislate.

UK prime minister Keir Starmer has been edging toward a social media ban for under-16s; now with added pressure from not only the Lords but also the Conservative Party leader, Kemi Badenoch, and 61 MPs sent an open letter supporting a ban like the one in Australia. Ofcom reports that 22% of children aged eight to 17 have a false user age of over 18 – but also that often it’s with their parents’ help. Would this be different under a national ban?

Starmer reportedly wants to delay deciding until evidence from Australia and, one presumes, from the consultation, is available. A sensible idea we hope is not doomed to failure.

Illustrations: Time magazine’s 1995 “Cyberporn” cover, which raised early alarm about kids online. Based on a fraudulent study, it nonetheless influenced policy-making for some years.

Also this week:
At the Plutopia podcast, we interview Dave Evans on his work to combat misinformation.

Wendy M. Grossman is an award-winning journalist. Her Web site has an extensive archive of her books, articles, and music, and an archive of earlier columns in this series. She is a contributing editor for the Plutopia News Network podcast. Follow on Mastodon or Bluesky.

Hallucinations

It makes obvious sense that the people most personally affected by a crime should have the right to present their views in court. Last week, in Arizona, Stacey Wales, the sister of Chris Pelkey, who was killed in a road rage shooting in 2021, delegated her victim impact statement offering forgiveness to Pelkey’s artificially-generated video likeness. According to Cy Neff at the Guardian, the judge praised this use of AI and said he felt the forgiveness was “genuine”. It is unknown if it affected his sentencing.

It feels instinctively wrong to use a synthesized likeness this way to represent living relatives, who could have written any script they chose – even, had they so desired, one presenting this reportedly peaceful religious man’s views as a fierce desire for vengeance. *Of course* seeing it acted out by a movie-like AI simulation of the deceased victim packs emotional punch. But that doesn’t make it *true* or, as Wales calls it at the YouTube video link above, “his own impact statement”. It remains the thoughts of his family and friends, culled from their possibly imperfect memories of things Pelkey said during his lifetime, and if it’s going to be presented in a court, it ought to be presented by the people who wrote the script.

This is especially true because humans are so susceptible to forming relationships with *anything*, whether it’s a basketball that reminds you of home, as in the 2000 movie Cast Away, or a chatbot that appears to answer your questions, as in 1966’s ELIZA or today’s ChatGPT.

There is a lot of that about. Recently, Miles Klee reported at Rolling Stone that numerous individuals are losing loved ones to “spiritual fantasies” engendered by intensive and deepening interaction with chatbots. This reminds of Ouija boards, which seem to respond to people’s questions but in reality react to small muscle movements in the operators’ hands.

Ouija boards “lie” because their operators unconsciously guide them to spell out words via the ideomotor effect. Those small, unnoticed muscle movements are also, more impressively, responsible for table tilting. The operators add to the illusion by interpreting the meaning of whatever the Ouija board spells out.

Chatbots “hallucinate” because the underlying large language models, based on math and statistics, predict the most likely next words and phrases with no understanding of meaning. But a conundrum is developing: as the large language models underlying chatbots improve, the bots are becoming *more*, not less, prone to deliver untruths.

At The Register, Thomas Claburn reports that researchers at Carnegie-Mellon, the University of Michigan, and the Allen Institute for AI find that AI models will “lie” in to order to meet the goals set for them. In the example in their paper, a chatbot instructed to sell a new painkiller that the company knows is more addictive than its predecessor will deny its addictiveness in the interests of making the sale. This is where who owns the technology and sets its parameters is crucial.

This result shouldn’t be too surprising. In her 2019 book, You Look Like a Thing and I Love You, Janelle Shane highlighted AIs’ tendency to come up with “short-cuts” that defy human expectations and limitations to achieve the goals set for them. No one has yet reported that a chatbot has been intentionally programmed to lead its users from simple scheduling to a belief that they are talking to a god – or are one themselves, as Klee reports. This seems more like operator error, as unconscious as the ideomotor effect

OpenAI reported at the end of April that it was rolling back GPT-4o to an earlier version because the chatbot had become too “sycophantic”. Tthe chatbot’s tendency to flatter its users apparently derived from the company’s attempt to make it “feel more intuitive”.

It’s less clear why Elon Musk’s Grok has been shoehorning rants alleging white genocide in South Africa into every answer it gives to every question, no matter how unrelated, as Kyle Orland reports at Ars Technica.

Meanwhile, at the New York Times Cade Metz and Karen Weise find that AI hallucinations are getting worse as the bots become more powerful. They give examples, but we all have our own: irrelevant search results, flat-out wrong information, made-up legal citations. Metz and Weise say “it’s not entirely clear why”, but note that the reasoning systems that DeepSeek so explosively introduced in February are more prone to errors, and that those errors compound the more time they spend stepping through a problem. That seems logical, just as a tiny error in an early step can completely derail a mathematical proof.

This all being the case, it would be nice if people would pause to rethink how they use this technology. At Lawfare, Cullen O’Keefe and Ketan Ramakrishnan are already warning about the next stage, agentic AI, which is being touted as a way to automate law enforcement. Lacking fear of punishment, AIs don’t have the motivations humans do to follow the law (nor can a mistargeted individual reason with them). Therefore, they must be instructed to follow the law, with all the problems of translating human legal code into binary code that implies.

I miss so much the days when you could chat online with a machine and know that really underneath it was just a human playing pranks.

Illustrations: “Mystic Tray” Ouija board (via Wikimedia).

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. She is a contributing editor for the Plutopia News Network podcast. Follow on Mastodon or Bluesky.