[simonw]

"AI really isn't a skill that needs to be taught"

At this point that's like saying Microsoft Excel isn't a skill that needs to be taught.

None of this stuff is easy to use, or obvious. If you want to get meaningful results out of it and avoid the many, many traps then there is an absolute ton you need to learn.

In the context of this conversation, the skill that needs to be learned is how to use AI to learn effectively. That gets into pedagogy and personal learning styles and self-discipline and all sorts of other extremely gnarly areas.

[operatingthetan]

Isn't the industry's idea that 'prompt engineering' is over and anyone can use this stuff effectively?

There seems to be a literal trap where people are too trusting of the LLM and take its word on code or whatever is being offered instead of reading it themselves.

In the context of the classroom this means teaching discernment more than ever.

[breezybottom]

I don't like the conflation of terms, but they're obviously referring to LLMs, which are designed to use natural language prompts. And the integration of LLMs into Excel means that to some degree Excel doesn't need to be taught anymore.

[marginalia_nu]

Well let's say that is true, and we do teach a class of kids how to use AI, how much of that will still be applicable by the time they enter the work force?

[simonw]

The bits about learning to communicate in clear and unambiguous way, having a skeptical ear and knowing when to double-check information that is being presented to them, and understanding how to use tools in a way that enhance their own learning should ideally last a lifetime.

[Avshalom]

Communicating in a clear unambiguous way doesn't actually help with using llms.

[simonw]

Disagree. Many people are really bad at clear communication. It doesn't matter how good an LLM gets, it will still be less effective to use it if you give it prompts with no single obvious meaning - you'll have to go a few extra rounds with it asking clarifying questions, or it will go ahead and do a thing you didn't mean it to do because your instructions could be reasonably interpreted another way.