Bryson reframes AI 'relationships' as authorship, not friendship
TL;DR
- Bryson argues the only real link between humans and AI is authorship, between the moral agent who built the artifact and the artifact.
- She describes AI as 'an extension of corporate intelligence' and 'not itself an entity we could be friends with'.
- Bryson still concedes humans will be affected implicitly by anthropomorphised AI, even when told the counterpart is corporate.
Professor Joanna Bryson has been making the same argument about chatbots for years, and this latest YouTube conversation restates it cleanly. Her position is that AI is an artifact, not a party to a relationship, and when people describe being 'in a relationship' with a chatbot they are describing something projected back onto themselves through a product a company built and owns.
The frame she keeps returning to is authorship. On her own blog she wrote that 'there's only one real relationship between AI & humans, that of the moral agents to their artefacts, the relationship of authorship', and then conceded the catch in her own argument, that 'most people are not authoring their AI at all', they are meeting it as an other. In an interview with Carnegie Council she puts the same point more bluntly, saying AI is 'an extension of corporate intelligence' and 'not itself an entity we could be friends with'.
Why this matters if you ship or regulate consumer AI: when a vendor sells 'companion' or 'relationship' framing over a chat interface, Bryson's read is that 'we may be relegating humans down to the level of machines' and hiding the corporate agency behind the character. The counterpart in the interaction is the company, not the model. That shift in language changes who is responsible when an emotional interaction goes badly, and it changes what a regulator can plausibly ask a vendor to disclose.
The honest caveat is that this is a normative argument more than an empirical finding. Bryson herself concedes we are 'programmed by biology such that we will necessarily be affected implicitly by anthropomorphised AI', so telling users to feel differently is not a real intervention. What the reporting here does not give you is any concrete policy proposal, or any sign that vendors will voluntarily strip the companion language from their marketing. The people who most benefit from her framing are the regulators, journalists and product leads who want a clean shorthand for an intuition they already had: the AI is an artifact, the accountable party is whoever shipped it, treat it that way.
Shared on Bluesky by 2 AI experts
-
I wondered what had happened to this. If it looks like I'm hot, I am :-) www.youtube.com/watch?v=IIhb... random podcast, but better than average questions.
View on Bluesky →
Originally reported by youtube.com
Read the original article →Original headline: Professor Joanna Bryson: "You Don't Have a Relationship With AI, You Have One With Yourself"