Safety
The current AI and I suspect that all AI of the near future (few human generations that is) is in fact a more sophisticated form of data acc…
The current AI and I suspect that all AI of the near future (few human generations that is) is in fact a more sophisticated form of data access methods in the same category as classic database indexes
The current AI and I suspect that all AI of the near future (few human generations that is) is in fact a more sophisticated form of data access methods in the same category as classic database indexes but with a lot more complexity and a statistical aproximation core. It also bundles innately a human language parser and generator. A pretty powerful new kind of data acces tool based on statistical aproximation that can closely mimic intelligence yet never really reaching the genuine artifact. People with shallow understanding of this architecture are easily fooled to think that if the machine is "good" a x & y (usually because of specific training methods) then it must also he goid at z,q & w abilities. The bias is a confirmation/hallo one and it is so ingrained that it become belief and a techno-religion now. Human abilities to imagine contains different abilities rooted in how our consciousness is participating in shaping reality. I know, this sounds like yet another religious view, but on a closer look you'll see that it is based on science. I don't think that the carbon biological substrate is the only one to hold a consciousness but the current AI is more like a veil shrouding the real and so aproximating it closely yet never reaching identity with the original. When the error level goes under our native abilities to detect we are caught in a mirage that makes us unable to distinguish fake from genuine. This doesn't mean that these AI tools are useless, or will kill us all trough their own initiative. What it means is that we are on the precipice of fooling ourselves and inflict self harm via these complex tools we are utterly unprepared to use safely. The "slow down" isn't about thinking slower or dropping entirely the technology but rather divert our efforts from the current mindless scalling to a more mindful process. A process that must include also explaining and teach how AI really works and safety use to all non-technical people, which is 99.999% (guess) of us. 🙏🥂🖖🇨🇦
Related
- the issue here is not that current or near-future machines are somehow “too smart”; it’s that they are too unreliable, and hence not worthy …
- i suspect that this is a wild underestimate, both ignoring the costs in developing the model and ignoring the fact that many questions may w…
- https://www.irozhlas.cz/zivotni-styl/spolecnost/marcus-jazykove-modely-ve-skutecnosti-inteligentni-nejsou-od-lidskeho-mozku-se_2606201209_ak…
- This was right five years ago, and still is: “Large scale pretrained models are certainly likely to figure prominently in artificial intelli…
Source: Gary Marcus (X) | 2026-09-15