#AIethics in the enterprise was a low interest rate phenomenon. In 2023, the stakes are too high, and developing powerful AI to own the future of search / the web is the 'win.' In 2019, political signaling to win talent was the 'win.' I'm not saying 'AI ethics shouldn't exist', I'm saying that you have to be childishly naive to think that companies, competing in a harsh environment with powerful adversaries, with huge monthly expenses, would hamper themselves with anything at all that didn't help them WIN. The level of naivety of those who thought anything else is absolutely embarrassing. You want to attract Stanford grads with blue hair, you gotta do it. If we keep catering to the guard rails and try to make anything perfect before we are looking at the disruptively phenomenal possibilities - it might be - as you say - a delay - that even though unquantifiable - would seem the wrong decision and thus not worthy as the path to choose.ĪI ethics teams were created to compete for talent and good-guy perception points. We have got to get more comfortable over-indexing on the positives of technological advancements because the nature of social and media will allow the negative news to become pervasive pretty quickly - hence under-indexing on the net benefits. That speaks to the nature of the next decade.

It is so uncertain - the quantification - and yet it is not a moot point you raise even if it is unquantifiable. Historically we haven't known these nor have we looked at them in such a way. We also don't know the cost of the advancements or accelerating them either - when it comes to AI specially. We don't know because it is hard to calculate. You make a great point on 'what is the actual human cost of delaying any advancement?'.

Reid hoffman linkedin