25% of googles code has been generated by Ai and checked by staff.
So this might be a marketing piece to bounce the share price up or hold it in place, or trying to make google staff feel valued whilst they take a pay cut or endure no payrise but…
If the results that google show me for results pertaining to code are anything to go by, I am not impressed. I cant even pass the I’m not a robot test sometimes and when I do, it still passes with wrong squares ticked!
Ive not been able to use a LLM to generate code yet, psuedo code, a few lines but its not something I can rely on.
Secondly, if search results can be poisoned en-mass which is what I see, how long before these LLM’s become poisoned?
Ive not seen the safeguards in the search results and not seen them in the LLM’s.
Theres also another problem, app stores couldnt handle an influx of new Ai generated apps and I dont see any improvements in underlying OS’es although thats more levelled at MS, but google chrome books can still be included.
This shows how easy it is to infect new malware for example. Only 2 AV products detected it, and thats only because the 2 AV’s detected 3 and 5 files being opened and encrypted respectively.
Even then whilst classed as a pass, it was still a blunt test that got around those behavioural hueristic detection algo’s, proper malware is more stealthy than that.
Anyway, I still think the templates have better mileage, but for some reason only a few programmers seem to understand them or their potential.
Perhaps thats an indication of the quality of coder thats around today which perhaps devalues google employees?