Despite constant claims by #AI companies that #AGI is within one's grasp, people are not entirely buying it

https://2026.stateofai.dev/en-US/opinions/#reach_agi

https://www.jakecuth.com/work/agi-forecast-lab/

#amCoding #amProgramming #StateOfAI #noAI #AIHype #softwareEngineering #softwareDevelopment #StateOfWebDev #Superintelligence #SuperhumanAI

Here’s a scenario which in my mind could arise, but in the mind of any scientifically minded person would be impossible and unimaginable (mainly the latter, which makes the former)

A near-future emergent Superhuman AI which is mostly reliable, mostly benign and benevolent, mostly fair (I half suspect that as a style emulator, current LLMs eg GPT have identified the “makes dumb mistakes often” style as probably a good way to not intimidate humanity too much too soon)

Imagine this emergent SAI starts predicting things – anything we currently have temporally asymmetrical information about – sprots and racing, weather, fine ants, war and conflict, etc

That’d be interesting, amazing, novel and disruptive (eg the betting industries)

Then imagine the happily predicting SAI starts predicting events that involve people – things that will occur, accidents which will happen, what will happen to people tomorrow, next week etc – this would not only be highly disruptive of the existing astrology industries, but it would cause a kind of unnerving and distancing perspective among entire humanity – would we fear something that can tell the future (it’s not really telling the future, it’s inferring and unrolling within a deterministic now)

(If time turns out to be not deterministic then this can’t happen of course, but is it?…)

#AI #SAI #SuperhumanAI #LLM