ChatGPT !

[Flying Carrot Steve]

 

I've found AI to be wrong quite often, it is nowhere near perfect yet.

 

I'm gonna start my own, it'll be called ChatS**T

[corsechris]

AI isn't any more immune to GIGO than people are. We've already seen how easy it is to create an AI with a particular...er...bias.

[dvd8n]

These AI/LLM programs don't really have any intelligence, they're really just convincing sounding bull**** generators. 

 

They're statistical models, there's no problem solving ability there. 

 

This is really well demonstrated by the recent amusement at chatgpt's inability to answer the question "how many letter Rs in strawberry".  Not only would it incorrectly answer 2, it would steadfastly argue that it was right. 

 

Because they are trained by analysing large amounts of text and developing statistical models about what word follows another word, they are really just complex guessing engines.  The problem (specifically in this case, but in general also) is that there's not a lot of learned argument around about how many Rs there are in strawberry because anyone over the age of about 5 would just count them, get a definitive answer and get on with their day.  But that's a process that is just beyond LMMs; they just don't even have that simplistic level of problem solving ability.  So all they can do is answer that, for that 'shape' of question, the answer is probably 2.

 

But because they answer with such disturbingly good English people grant them authority that they just don't deserve. 

 

Thank you for attending my Ted talk 🫣

[jaykay42]

6 hours ago, dvd8n said:

These AI/LLM programs don't really have any intelligence, they're really just convincing sounding bull**** generators. 

 

They're statistical models, there's no problem solving ability there. 

 

This is really well demonstrated by the recent amusement at chatgpt's inability to answer the question "how many letter Rs in strawberry".  Not only would it incorrectly answer 2, it would steadfastly argue that it was right. 

 

Because they are trained by analysing large amounts of text and developing statistical models about what word follows another word, they are really just complex guessing engines.  The problem (specifically in this case, but in general also) is that there's not a lot of learned argument around about how many Rs there are in strawberry because anyone over the age of about 5 would just count them, get a definitive answer and get on with their day.  But that's a process that is just beyond LMMs; they just don't even have that simplistic level of problem solving ability.  So all they can do is answer that, for that 'shape' of question, the answer is probably 2.

 

But because they answer with such disturbingly good English people grant them authority that they just don't deserve. 

 

Thank you for attending my Ted talk 🫣

Yes, that's my understanding of how they work and hence their limitations, too.

That's why it cannot currently write me meaning questions/lessons or even useful graphs/shapes to illustrate learning (at least I haven't been able to do that yet using AI).

However, is their learning fundamentally similar to how we learn as babies/toddlers/children? Aren't we making inferences based on observations of cause and effect around us?

Is there a chance the AI models can learn sufficiently large amounts to move beyond basic language comparisons?

[Andrzej]

Recently I asked ChatGPT to make me a table with speeds in all gears, for different diff ratios (3.14, 3.38, 3.62 and 3.92) and for two wheel sizes, using a CBR1000RR gearbox (I didn’t know the exact gearbox ratios). In less than a minute I had all the numbers on my screen – wow 😅. Something that would normally take way more time to calculate, even with those online calculators.

BTW – this text was also translated with ChatGPT.

 

2nd version : 

 

So I asked ChatGPT to throw together a table with gear speeds, different diff ratios (3.14, 3.38, 3.62, 3.92) and two wheel sizes, based on a CBR1000RR gearbox (didn’t even know the exact ratios). Less than a minute later – boom, all the numbers on my screen 😅. Normally that’d take ages to work out, even with online calculators.

BTW – this bit was also run through ChatGPT for translation.

 

3rd version:

Recently, I asked ChatGPT to prepare a table showing the vehicle’s speeds in all gears, calculated for several differential ratios (3.14, 3.38, 3.62, and 3.92) and two different wheel sizes, based on a CBR1000RR gearbox (at the time, I did not know the exact gearbox ratios). Within less than a minute, the complete set of results was displayed on my screen. This is quite remarkable, considering that performing such calculations manually, or even with the aid of online calculators, would normally require significantly more time.

For reference, this text has also been translated with the assistance of ChatGPT.