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!
From grade schools to advanced degrees will AI/ML become the ultimate teaching tool? Will it adapt constantly to improve its time to teach and quality?
The potential to improve teaching in grade schools to advanced engineering and customer service will be only limited by the imagination. Could it also vastly lower the cost of teaching by not having books and used to teach, but also tutor those that need it? I recently interfaced with an ATT AI/ML to program new remotes. A few bugs, but it indicates the future use, especially it can modify itself to improve its results on the fly.
AI/ML has the ability to improve teaching, use and repair almost everything as it replaced obsolete learning, teaching and repair almost everything. I see the ability of AI/ML/teaching to learn and advance itself on the fly to be the next real game changer. Any thoughts or additions sought and welcome for this is the ultimate ecosystem for it almost resembles a living thing.
This seems to be in part predicated on the idea that AI/ML systems will actually reliably learn and improve as they go - and I think it's also implied this will happen without much (if any) expensive (and occasionally error prone) human interaction. Should we imagine this as a stable, closed loop system ? I'm also reminded of the Vince Lombardi corollary: "perfect practice makes perfect" - i.e. practicing errors generates more errors. When will we have enough confidence that AI is doing perfect practice ?
Humans are still quite involved in the design of complex systems today (AI chips and infrastructure, EDA tools and flows, etc). Do we really foresee a day when AI/ML can reliably check and correct itself and human checking and debugging skills are no longer needed ?
There are certainly layers in the "learning stack" that could largely be replaced by AI/ML and as the depth and breadth of knowledge required to master any given technical area expands, it's arguable this is a good thing. My concern is that once you lose the "full stack" understanding you can't actually debug the final system any more. Recent experience suggests - for example - that today's students trust digital measuring equipment far more than they should and they often confuse meter resolution for accuracy. It is all too easy to slip into a world of illusory precision.
For me, a lot of what makes a good engineer is having a good feel for how things work and what sort of numbers and behaviour are in spec and which are suspect. If AI can either do that reliably or improve the competence of the humans still required, that would be progress in my book. But I'm not close enough to what's happening to know if that's already happening or likely to.
From grade schools to advanced degrees will AI/ML become the ultimate teaching tool? Will it adapt constantly to improve its time to teach and quality?
The potential to improve teaching in grade schools to advanced engineering and customer service will be only limited by the imagination. Could it also vastly lower the cost of teaching by not having books and used to teach, but also tutor those that need it? I recently interfaced with an ATT AI/ML to program new remotes. A few bugs, but it indicates the future use, especially it can modify itself to improve its results on the fly.
AI/ML has the ability to improve teaching, use and repair almost everything as it replaced obsolete learning, teaching and repair almost everything. I see the ability of AI/ML/teaching to learn and advance itself on the fly to be the next real game changer. Any thoughts or additions sought and welcome for this is the ultimate ecosystem for it almost resembles a living thing.
AI already is the ultimate learning tool. I started grad school when web search just started. I learned much more using search and saved an amazing amount of time. I graduated with honors even, since the rest of the class did not use search yet or even had personal laptops. I had a DG1 if you remember back then.
Ai will be 100x search for learning. As I said, if you are not using AI today you are already behind. AI results now come with disclaimers so even the lawyers are in on it!
