If you look at the actual ranges on his graphs, the conclusion is a little bit exaggerated, but the basic point is valid.
Education systems need to have very strong curriculum about AI use and the necessity of just not using it to some degree to be able to obtain any level of actual literacy or other competency.
Actually education should be redesigned to properly incorporate personalized AI tutoring and as part of that effort provide a supportive environment where students don't feel the need to go around it to cheat.
If the goal is to communicate and solve problems, the technologies will probably be a huge improvement if we can manage them properly.
I definitely do not have the depth of vocabulary that some 18th century readers had, but I am a good problem solver. Probably a more effective problem solver with AI. But kids do need to be trained that they will lose it if they don't use it.
If you look at the extreme amount of video content these days and combine that with the increasing abilities of AI video generation, there may be a trend towards more visual (and often more literal) communication.
I often find that screenshots or screencasts are important for technical communication.
But of course we don't have any visual replacement for the abilities of natural language so I hope we can keep that.
We need a deliberate and effective effort in education.
But we are rapidly approaching the era where your access to AI and robotics determines your labor productivity. So it increases the existing inequalities.
It comes down to the topology of the social networks and how the built environment and belief systems shape that. But the belief systems mostly serve the social groups rather than the other way around. So maybe what happens is determined by group dynamics.
Thank you to anyone who read my ramble.
> we are rapidly approaching the era where your access to AI and robotics determines your labor productivity
Some of the wealthiest, most powerful and yes, most productive people of the last decades had no clue how to use a computer or phone, interacting with it through staff.
Remember the folks who were teaching their kids to code ten years ago? How relevant is that today (beyond as a cognitive exercise)? Access to AI and robotics are secondary to other factors. Not the determinative ones, certainly not at the individual level.
Please parse that sentence more carefully. I said "labor productivity".
> Please parse that sentence more carefully. I said "labor productivity"
Yes. Unless we’re redefining labour to fit a square peg into a round hole, this remains true.
Access to the latest technology is never a sole determinant of productivity in any technological revolution. If anything, it having any relation to productivity is a modern phenomenon.
(Most kids who had a computer in the 80s or internet in the 90s didn’t become wildly more productive for it. They just found new ways to entertain themselves.)
> Actually education should be redesigned to properly incorporate personalized AI tutoring and as part of that effort provide a supportive environment where students don't feel the need to go around it to cheat.
I have no idea what is this magical AI capable of tutoring that you're talking about. It is certainly not any of the AI models we see in the market, which hallucinate systematically and are only able to output remotely valid content if they are subjected to tight feedback loops.
I have built such a tutoring system.
This is another magical thinking we have in our age. AI will do nothing good for education, because to be educated you need to do the hard work of understanding things. This is just the opposite of getting a machine to provide you with quick facts, summaries, and conclusions. In fact I would say that if you introduce AI to classrooms it would be better to take a good number of $100 bills and incinerate, because the end result will be less expensive.
Can you try to read my comment more carefully?