Thread by @j2blather

Probably won't make it all the way through the last session, need to go back to the hotel and nurse my cold :-/ hanging out though in the back to see how Emma Kendrew is doing to talk about. #m3london

As usual, @Accenture seems clued in about AI. Kendrew talking about how people don't know they are using AI when they use google, spotify, etc. Whoops, now talking about it "as a coworker", I obviously object to that language!

Kendrew hows predictions by 2020 of 85% of customer interaction by chatbot, 2025 a government will ban human driven vehicles, next decade "an AI" will make a blockbuster film. [Well, I doubt all those predictions! RBS found chatbots included calls to the banks' humans!]

argh, Kendrew is totally anthropomorphising AI, claims it can learn independently, evolve independently "if we let it." This is a very, very dangerous framing, cf. link.springer.com/article/10.100… "raising responsible, citizen AI" need to train & teach, can't limit it. #m3London

Wow, I take back what I said about @accenture. Kendrew is talking about sensible things like bias, but in a completely unsensible and dangerous way. We need to maintain HUMAN accountability. Tieto has "an AI" on its "leadership team" theverge.com/2017/10/30/165… #aiethics #m3london

Kendrew: @united used decade-old data and lost $1B of business opportunities as an example. #m3london. @amazon is trying to "incentivise truth" on their reviews. Ridesharing drivers were coordinating surge pricing by coordinating signoffs to create a surge.

Kendrew is now saying that we need to augment & benefit humans / employees, but then goes back to the "raising purposeful AI" metaphor. #m3london success factors: scaling, UEX as BAU, scaling, continuous curation of data, monitor& manage, privacy @ scale.

Kendrew "monitoring & managing AI is different than any other tech" , this is exactly what both I and @lmarsden's talks were directly opposed to, AND most of the @APPG_AI / UK govt findings. We need to require companies using AI to conform to extant law. #m3london @accenture

Kendrew is preaching human-centricity well, but still is treating AI as anthropomorphised. Now says trust is a big problem. Emphasises transparency (which is great!), bias etc. laws, [but not #cybersecruity which is key]. We need to know its purpose. Closes calling #ai a tool!

Talking to Kendrew afterwards/q&a she shares a lot of my concerns on human accountability & just didn't see the anthropomorphism as problematic. I may have scared her too much, I'd just like to see the same talk reframed as ALMOST like raising, SOMEWHAT like coworkers. #m3london