Ah, sorry, I used to keep posting about the theme over here (although now I have stopped), so accidentally continued on that. But a brief summary in this set of responses (to save a bombardment of replies) x.com/_onionesque/... I find open peer review important for different structural reasons.

I am more interested in this not because of "peer review" directly, but because I think it is one very concrete aspect by which you could make a difference to a far more serious structural problem that keeps eroding the legitimacy of the whole enterprise.

The fully non-blind review process that I am aware of is by F1000 (which I do happen to like). But various venues that offer to opt-in typically wait till after the decisions have been made to make names public.

It will accelerate in about 2-3 years IMO -- when after sufficient trial and error, enough money burnt into smoke, orgs would have paid enough to earn the (non-tech) expertise.

Business adoption is only slowing because developing the right AI strategy will take time. Knowing how to use it, where to ise it, how to re-engineer a portion of your your org around a new co-pilot (for instance) for it to actually shine, how to implement change management, is the hard part.

The likely consequence of this might be a persistent underinvestment in foundational scientific, mathematical, and (core) engineering capacities. But I wonder how can that be quantified.

encourage generating deep human capital formation. It may have instead displaced it, replacing slow, cumulative intellectual development, with immediately monetizable competence (which seems to have lower transferability over time). [+]

One can see that it caused a massive demand shock that may have permanently altered the human capital trajectory of an entire set of cohorts. It has this interesting facet (which is often lampooned, correctly) that a lot of it appears "high-skill" on the surface, but clearly does not [+]

He seems to like the imprecise probability framings. But I like the idea that if you have some plausibility-based ideas of what the object at hand, you should be more efficient. Surely this can't be the first paper on the theme (beyond imprecise formulations)?

Still trying to understand some stuff around more general e-processes. Do you know of any generalizations that have a similar flavor?

Unrelatedly, another concrete (maybe niche-seeming) area seeing major disruptions is in revenue management. While there isn't any data on this, quite a few of my friends have mentioned how many customer success and junior analyst roles have been automated (this is fully believable/ML-attributable).

Note that the 70% figure is not readily attributable to AI. The world economy is in doldrums, and such jobs usually get very tight in such periods. But whatever the true percentage, the combined effects already seem to be re-shaping the whole industry (tighter budgets combined with more tools).

Slightly more lucrative used to be in the video game industry. Not uniformly studied, but in some places (such as China) there were reports of up to 70% job losses restofworld.org/2023/ai-chin... two years ago. Also true for NGOs and for campaign illustration. content marketing etc.

Some of the most anti AI people I know IRL (very offline. Despite my frequency of posting, I am also not very online, barely lurk, just post in bursts and sign out) happen to be illustrators. A large bulk of their work was 'commodity' illustration e.g. like on mugs, for websites etc.

For instance, take stock illustrations in magazines and newspapers now. They are far less likely to be credited. While for sure someone reviewed and created it, it was more likely to be created by AI tools (this needs a study to quantify the real size of effects, since it's easily measurable).

Writing is a different ball game. Logo designers, illustrators overwhelmingly freelance. It's also about shaping the idea, but the relation is a bit different. A large amount of work is for stock illustrations and concept figures, which are easier to offload. I had read 40% reduction in job postings

Illustration and graphics design markets are quite disrupted. They were moderately growing a few years ago, and now starting to completely crater in the past two years.

*can you get any-time validity