I won’t get into tit for tat internet debates anymore. Everything is conjecture here, we’re trying to forecast the future. That said, I was a BIG SKEPTIC until a couple of months ago. Tech changes quickly. Whatever your opinion of Ai, or any advanced tech, re evaluate it every 6 months. I manage data science projects for a living, if that’s worth anything. There’s way too hype, but I caution against opposite but equal hubris.
This will be unpopular, but he’s almost right. Most white collar jobs could be automated to the point where minimal human oversight is needed, with technology we’ve had for a decade. It hasn’t happened because developing that software has been prohibitively expensive.
What has changed is the cost of developing software for the “last mile” (logistics parlance) is dropping like a stone thanks to AI. This isn’t entirely about AI doing your job, but AI analyzing your job and finally using normal procedural tech to automate it. Your role may be reduced to supervision.
For jobs that need actual cognitive power, the quasi-reasoning capability of LLMs will likely supplement your work, if it’s not already. It’s unlikely that high stakes reasoning will ever be 100% automated, but surely that’s a judgement of the responsible party or more likely the irresponsible party.
This will create social upheaval and psychological torment for many. The obvious economic disaster of massive unemployment of educated people who are used to doing highly organized functions is making some people wring their hands. Certainly there should be a plan, but capital insists a race to the bottom immediately or risk being obsolete and worthless. First movers may cash out before everyone else loses their shirt.
Our engineers can use Linux desktop if they want, and I suppose anyone else could as well, but Microsoft Office is really what keeps me on Windows at work. I could use the browser based apps for 80% but that last 20% is nasty. And yes, I use libreoffice at home. The cross compatibility just isn’t there without loads of extra time that I don’t have.
It’s OK to hate AI slop and recognize the immediate threat to cyber security it brings. At least they are trying to mitigate it. There’s been no similar actions from other frontier models. They are deliberately helping open source projects with little funding to keep pace.
https://www.anthropic.com/glasswing