Quote:
Originally Posted by marsplasticeraser
It's funny, as we were just doing some AI testing around facts.
AI sounds very definitive, but in about 25% of the cases gave completely incorrect facts.
It reminds me of that guy who emphatically tells you about incorrect facts, loudly.
The one I remember was about US salary data. It got the mean right (I think that was around $60k), but had median salary around $80k. Like it was so obviously wrong that it's kind of surprising, as there is no nuance in getting that fact.
The issue with AI is they don't give you the backing data to interpret if what it says is correct. If you like data and understand how much variety there is in a 'fact', this is a huge problem.
My take: AI is good for people that simply want an answer and aren't focused if the answer is correct. AI is really flawed when it comes to nuance and things that are grey, though it is ok at presenting concepts.
|
AI is basically like a loud user on Reddit. Makes sense since it gets a lot of data from Reddit.