AI experiences, GPT3 Leta series

[Rollo62 538]

Hi there,

 

I received my OpenAI beta test account recently, after waiting for that more than a year now.

Finally I was very much excited about the AI capabilities, as I watched the amazing Youtube series of GPT3 Leta from Dr. Alan Thompson,
who is experimenting with the DaVinci model and avatars to provide a real human chat experience.

This machine will easily pass the turing test IMHO.

 

Well, but when I first checked the Playground, I found the GPT3 models were not so much evolved as I expected.

I also tried out the REST API, with the same disapointing outcome.

 

GPT3 often gives completely wrong answers, mostly doesn't understand anything at all, not even a name or very basic texts,
and all in all its far from capabilities of the Leta series I mentioned above.

So maybe this is the case because I have not provided enough or wrong training data, but there is also no real explanation howto provide information,

and especially what data would be really needed, to enable GPT3 to deliver most useful results.

 

In the meantime during waiting for GPT3 access, I already tried GPT-J and tried others, but they all were not especially "intelligent" neither in understanding normal texts

nor in providing real world answers (like: who won the latest nobel price in physics ?).

Now I would say GPT3 is on a same level as GPT-J, at least when it comes to a unspecific chat application.

What is Dr. Thompson's secret to make it so clever ?

I assume that all GPT3 models use the same training data ( it was mentioned that training data was used only from the time until year 2018 ),

so that also Dr. Thompson's machine should be using the same data.

 

I wanted to ask if anybody here in DP had also some AI experiences, and maybe could squeeze better results out of GPT3 ?

Would be great to exchange some tips and hints. howto handle AI best.

 

[Guest]

Why would you dabble with that?

Do you really see a future non-problematic increase in income?

I am seriously interested in your use case.

The GPTs and "AI"s in common are laden with problems and pitfalls both technical and political.

Edited September 23, 2021 by Guest

[Rollo62 538]

1 hour ago, Dany Marmur said:

Why would you dabble with that?

Do you really see a future non-problematic increase in income?

I am seriously interested in your use case.

The GPTs and "AI"s in common are laden with problems and pitfalls both technical and political.

Don't you see any use-case, I wonder ?

 

If the a device chat can give perfect answers, like the Leta project can do, I see a lot.

Just as example for assisting people in their daily worksite, in real language, maybe by a special trainned AI which is optimized to the special field of the customers current working tasks.

The AI could act as a real useful assistant, giving useful tips.

 

But that is only one of many ideas, for the others I better let you sign an NDA

No not really, I think there have been already millions of people thought about the same ideas, like always if a new technology is emerging.

While OpenAI, Apple, Facebook and those kind of companies ahve already filed their patents.

 

"... increase in income ..."

Not sure how you mean that, but I think:

Yes, such AI gadgets in apps and software will sell veeeery well to a lot of company and private customers in the near future, if the AI can really do a good job.

 

 

[Alexander Sviridenkov 360]

In my country all mobile providers and banks use such AI, and all these AIs are so stupid that first thing that everyone do when calling is trying to force him to switch to human.

Just another annoying thing like long voice menus.

[Rollo62 538]

I absolutely agree, I have Alexa, Siri (only Google Assistant is more clever ), but they all are miles away from this Leta series.

 

I am not talking about those, but the GPT3 (Leta) videos in the link above, which is in an experimental, closed beta.

Are the Leta videos then a hoax finally, I dont think so ?

There are about 24 videos right now, and I think Dr. Thompson is close to many AI research teams, having a deeper insight as we could have.

Since GPT3 is only a textual processor, maybe the Q/A is somewhat "beautified" before sending to the avatar and speech output.

But many of Leta's answers are really astonishing, and would make AI really useful.

Edited September 24, 2021 by Rollo62

[Fr0sT.Brutal 900]

17 hours ago, Alexander Sviridenkov said:

In my country all mobile providers and banks use such AI, and all these AIs are so stupid that first thing that everyone do when calling is trying to force him to switch to human.

Just another annoying thing like long voice menus.

Nevertheless they're quite strong at defending human operators from contacting with clients. Sometimes "Give me alive operator you stupid bot!!!" phrase is not enough and you have to find more clever ways to break through this wall.

Edited September 24, 2021 by Fr0sT.Brutal