Artificial Intimacy News #4
A field report from the place where 21st Century technology meets human behaviour, culture, and evolution.
In this issue:
A delightful conference
People are gaming self-driving cars
Taking Snapchat and TikTok to court on behalf of young people
Anthropic is about to go voice-mode
This Week
I am attending the European Human Behaviour & Evolution conference (EHBEA 2025) at Northumbria University, Newcastle-Upon-Tyne. I am delighted at the little corner of Newcastle between my hotel and the conference venue, full of old sandstone buildings, castles, and an elegant cathedral. Despite the hilly terrain, there is a great track along the river Tyne, and, with the help of a whopping case of jetlag, I’ve managed to get out early for some tremendous runs.
On Tuesday, I spoke about some of my lab’s work using chatbots to understand both human-computer interaction and human-human social behaviour. A highlight of the conference is the attention to how AI is altering, abetting, or teaching us more about human behaviour. My colleagues are on board, and the 2026 conference, to be held in Leiden, Netherlands, will have an AI theme in recognition of perhaps the most important developing front in the behavioural sciences.
Insight
What happens when AI gets involved in activities where humans adopt behavioural conventions? We are just beginning to see, with the roll-out of AI-driven vehicles. Human drivers observe (most of the time) conventions like what side of the road to drive on, when to let others merge, and what to do at 4-way Stops.
, creator of the excellent Modern Wisdom podcast and newsletter, observes an interesting phenomenon emerging as Waymo rolls out its driverless taxis in Austin, TX, where he is based.I often write about how surprised I am at how willing people are to treat AIs – especially conversational AIs – very much as they would humans. Chris’s observations tell a different tale, one of humans choosing not to cooperate with driverless vehicles, knowing they can take advantage of the vehicles’ safety-first AI.
In other news, Waymo the driverless taxi service has come to Austin.
I’ve found over the last few weeks that a Waymo which says it’s 7 minutes away regularly takes 15 minutes to get to me.
It wasn’t getting lost or taking wrong turns, it just seemed to be moving unusually slowly around pedestrianised areas and junctions.
I’ve realised that Waymo is totally cucked and has taught me a fundamental truth about how humans think in cars.
Allow me to explain.
Humans largely behave on the road.
They do this to avoid inconveniencing or angering other drivers, and to avoid endangering themselves.
It’s reciprocal courtesy.
But with Waymo that isn’t a consideration because 1 - you know the Waymo will always stop if you bully it by crossing the street or pulling into its lane and 2 - there is no fear of retribution or sense of guilt for inconveniencing another driver as there is no driver.
So Waymo just gets bitched by every human who’s prepared to prioritise themselves.
Hence, it’s the last priority at every junction and crosswalk.
This issue will disappear once all vehicles are autonomous, because the playing field will become level again.
But I think the next few decades are actually going to see driverless vehicles be way slower to reach their destination until either humans stop putting themselves first on the road (never going to happen), or self-driving is given some sort of retributive algorithm to tailgate a driver who just cut you off.
Headlines & Highlights
The Verge - Anthropic is reportedly launching a voice AI you can speak to
Apparently we will have voice mode for the popular Claude LLM sometime this month. I wrote last year about the implications of OpenAI’s ChatGPT voice mode.
Academic Perspective on Artificial Intimacy
Delighted to see academic discourse about Artificial intimacy and its implications ramping up. For a quick read, check out “Psychologists explore ethical issues associated with human-AI relationships” the Cell Press story about a recent Trends in Cognitive Sciences perspective piece led Daniel Shank.
Lawsuits against TikTok and Now Snapchat
Jonathan Haidt and
at draw attention to the evidence that Snapchat is damaging teens, and that Snap.Inc knows about the massive scale and extent of this damage. Follows on from an earlier article about the wide-scale harms being caused by TikTok.
