Our hallucinating AI overlords

LLMs hallucinating information is a pretty common issue. That’s because they don’t contain knowledge at all, they are just good at predicting how to put words one after the next. This AI hype is so overblown, I’m pretty frustrated by all of it. If you guys are curious I found this video pretty interesting:

Ehh I tried watching the video and it seems to be mostly about Big Data and presumably says how the AI trend is similar to that.

And don't get me wrong, I'm sure 95% of the new "AI" startups could be looked at like a hoax similarly to Big Data. But then that's just a symptom of the VC system in general and the same could be said for any number of other IT companies in the US.

I know a lot of programmers who use AI in their daily work - not constantly and it won't code programs for you, but it's already a very useful tool and it has progressed massively in a short time. I'm not a programmer but ChatGPT created a python script for me which did what I needed perfectly, on the first try (couldn't find any ready made tools for that specific use).

Similarly we know companies use generative AI for marketing materials, and I personally use it for pedal graphics. It's also a useful tool if you work with text. It is just a tool and won't do amazing things perfectly, but ChatGPT for example is a great tool already if you know how to use it. The hype might be overblown, but LLM's have a huge impact already in the world and they're not going away.

(as a sidenote I am a bit worried about how future generations will fare as they can use ChatGPT for school work for example instead of learning, but then it's also hard to say how much of that is just the normal "the youngins are ruined! ruined I tell you!" that happens literally every generation)
 
Ehh I tried watching the video and it seems to be mostly about Big Data and presumably says how the AI trend is similar to that.

And don't get me wrong, I'm sure 95% of the new "AI" startups could be looked at like a hoax similarly to Big Data. But then that's just a symptom of the VC system in general and the same could be said for any number of other IT companies in the US.

I know a lot of programmers who use AI in their daily work - not constantly and it won't code programs for you, but it's already a very useful tool and it has progressed massively in a short time. I'm not a programmer but ChatGPT created a python script for me which did what I needed perfectly, on the first try (couldn't find any ready made tools for that specific use).

Similarly we know companies use generative AI for marketing materials, and I personally use it for pedal graphics. It's also a useful tool if you work with text. It is just a tool and won't do amazing things perfectly, but ChatGPT for example is a great tool already if you know how to use it. The hype might be overblown, but LLM's have a huge impact already in the world and they're not going away.

(as a sidenote I am a bit worried about how future generations will fare as they can use ChatGPT for school work for example instead of learning, but then it's also hard to say how much of that is just the normal "the youngins are ruined! ruined I tell you!" that happens literally every generation)
Oh yeah that video is not really about the tech. But it puts it in an interesting perspective. I use AI for coding every day and I’m not super impressed honestly, but I agree that LLMs are having an impact. It’s just silly to think that they are intelligent machines and I am skeptical that all the hype around what can be done with them will actually materialize in real products that are useful to us. Many of these companies are just riding the hype train. As somebody who works in tech I just find the environment a bit frustrating to deal with.
 
I use AI for coding every day and I’m not super impressed honestly, but I agree that LLMs are having an impact. It’s just silly to think that they are intelligent machines and I am skeptical that all the hype around what can be done with them will actually materialize in real products that are useful to us. Many of these companies are just riding the hype train. As somebody who works in tech I just find the environment a bit frustrating to deal with.
Right, AI (I still prefer the label 'machine learning') has legit uses when applied to the right problems, ones that are well defined and where good quality training data is available. But opening it up to unconstrained public use based upon training on wide-open internet scraping is just nuts in my view. It's often the case of garbage in, garbage out.
 
Right, AI (I still prefer the label 'machine learning') has legit uses when applied to the right problems, ones that are well defined and where good quality training data is available. But opening it up to unconstrained public use based upon training on wide-open internet scraping is just nuts in my view. It's often the case of garbage in, garbage out.
+1000 to that! ML good, AI bad :)
 
Seriously, we don’t have AI. We have super fast search engines designed to deliver info in a certain way. It only knows what we tell it to know. “AI” is a sham…they are trying to sell you something “new,” but it ain’t new. It’s a rip off. Innovation is dead because there’s no profit in it. Only true innovators like Josh Scott and that Ge diode guy know the truth…
 
The problem with AI for me is how the developers rip off writers, musicians, artists, etc. AI "music" is just an extension of sampling. To generate the "music" they sample millions of songs written by actual people and morph into something which sounds the same but different. Just like when people sample records to make a new tune. If you sample someone else's work you have to pay them, which is fair enough. You're using someone else's work. Why are the people who have used other people's work to develop their AI programs not paying for all the stuff they have used?
 
The problem with AI for me is how the developers rip off writers, musicians, artists, etc. AI "music" is just an extension of sampling. To generate the "music" they sample millions of songs written by actual people and morph into something which sounds the same but different. Just like when people sample records to make a new tune. If you sample someone else's work you have to pay them, which is fair enough. You're using someone else's work. Why are the people who have used other people's work to develop their AI programs not paying for all the stuff they have used?
Word.
 
FWIW I asked ChatGPT 4o and got this:
The word "Michigan" does not contain any instances of the letter "E". Therefore, the number of occurrences of the letter "E" in the word "Michigan" is zero.
I did ask it "How many of the letter E is there in the word "Michigan"?" which might make a difference.
 
Back
Top