You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
Will AI/ML/robotics lower demand for labor creating one of the greatest shifts in human history where even in a shrinking labor market can increase quality of life and wealth all without additional labor? If true, this will be one of the most dramatic changes in human history where technology allows even greater growth in a shrinking labor market?
Most certainly. It will also change the jobs it does not eliminate and hopefully create more jobs over time. The initial job loss shock will be significant however. I am already seeing customer support jobs taken over by AI. Marketing jobs are next. Most entry levels jobs can be AI automated. Coding jobs for sure. In the semiconductor industry 30% of the design cycle is already being "enhanced" by AI. That should be 50% once agentic AI hits production. Verification and debug will be hit hard by AI but I am also seeing AI layout tools and other physical implementation going AI. The more AI the less bugs/errors you will have.
Robotics and manufacturing are ready to collide resulting in reduced employment. Fabs are already automated but I am sure there is more to be done. Other manufacturing that is more antiquated will be targeted. QA is also a target. Less humans means less QA cycles.
I predicted AI will hit most job markets by 2030. It looks like it will be sooner. The people who are ranting about the employment gap, not having enough workers, will be shocked by double digit unemployment due to AI by 2030. Just my opinion of course.
I daresay Dan's take is representative of the FOMO stance of most high level management at this point in the AI adoption curve (or hype cycle curve if you prefer).
For those who have to implement those high level visions, the story is a little different. Especially when it comes to signing off on releasing to production the products created using AI. The folks at the 'coal face' are always trying to think whether they have covered all the ways something can go wrong. For them, AI presently represents a whole lotta possible new ways for things to go wrong. Which must be checked thoroughly.
Unhelpful during this AI push is the lack of real training to go with it. Even the big 3 eda vendors say, "just use this module, it works".* Engineers don't like 'black box' responses like that. They spend their entire careers taking apart things and rebuilding them to make them better. They can't do that with AI so they remain deeply suspicious.
*: Also there's the question, does this AI product work well enough, error free enough, to justify its price? My own limited experience with these products say 'no, not yet', and even Sam Altman has recently said AI will make up answers rather than admit it doesn't know.
The pressure at the top to adopt is real. But that doesn't make the high levels of risk go away.
