[supriyo-biswas]

The way I see it, generative AI has been introducing a lot of distrust into systems that worked "fine" previously, such as rendering homework ineffective in the case of education, making verification difficult for remote interviewing, flooding the internet with low-quality noise (aka slop) that makes it difficult for reputable and researched sources of information to stand out, with all the implications it has for society, the fraudulent returns described here, and the like.

Ultimately, it would be a bit ironic if generative AI ends up kneecaping itself, either through regulation (because businesses and governments will be unlikely to tolerate hiring fraud, returns fraud etc. beyond a threshold), or caused things to move into meatspace through on-site interviews, reliance on physical stores, elimination of online courses and others, which is less amenable to its application.

[thunky]

> rendering homework ineffective

Homework isn't any more ineffective imo. The way we educate and grade is.

Imagine if people went to school to learn something rather than to "level up". And you earned a job based on what you know, or what you can do, rather than what degree you banked.

Then maybe you would want to do the homework.

If gen AI helps us flip the current system on it's head that would be a good thing.

[intended]

No one is willing to pay for this. Bloom’s 2 sigma problem is from 1984.

MOOCs were the hope for education but that didn’t take off either. Now any remote learning will need physical examinations, which make certification pathways for everything more expensive.

Even if you want to study, our distractions are crafted by people who spend hours figuring out the right dopamine reward schedules to keep you distracted.

[thunky]

> Now any remote learning will need physical examination

I don't think so. Proving you can pass a test is pretty useless imo. Especially when it typically boils down to a memorization test and the subject matter is largely irrelevant.

[intended]

Proving you can pass an open book test ? Or doing graded assignments?

That wasn’t something possible prior to LLMs. Being able to cheat code or generate homework assignments was not this trivial either.

[thunky]

I think people need to figure out how to demonstrate they can add value on top of AI rather than showing they can do it all from scratch with a "closed book".

[intended]

This is at a level of abstraction that allows one to say something hopeful, and let the details blow up in one’s face for later.

A better example would be “we don’t use fork lifts to do our gym exercises”

The whole issue with AI is the verification labour. This requires people to have the skill / discernment to make out if the output is useless or not.

Building this human capability is fundamentally a product of learning, training and experience. All of which require exams, tests and verification.

[supriyo-biswas]

> And you earned a job based on what you know, or what you can do, rather than what degree you banked

As I see it, companies will want some sort of artifact that tells them that the person has some basic knowledge in their field of study, which would make us come back to the same system that many detest.

Otherwise we'll just end up with Samsung Korea's version of the entrance test[1] which is like the SAT but for getting a job, which only a handful of companies can realistically do, and as such there is very little appetite for it in the "West".

[1] https://www.graduatesfirst.com/samsung-aptitude-tests

[thunky]

> As I see it, companies will want some sort of artifact that tells them that the person has some basic knowledge in their field of study

Sure, but unfortunately a degree doesn't really do that today. When you interview someone, do you care at all about their degree or grades? Does it give you confidence they know something? I don't think so.

Which is my main point, that this isn't a new problem.

[carlosjobim]

Companies can do their own tests to find suitable candidates. They should.