[ValarMorghulis]

Satya Nadella hires Sam Altman to Microsoft along with the founder Greg Brockman.

Link

[DigitalOne]

There seems to be a clash between two school of thoughts.

One school of thought says AI is not inherently dangerous, can be hugely beneficial for humanity and should be developed within a for-profit corporate structures. Sam Altman, Satya Nadella/Microsoft/Google have this philosophy. T

The other school of thought is that AI can be dangerous to humanity, the development should be controlled/regulated, non-profits would be the best to do further research. Elon Musk, or remaining OpenAI board members*seem to tend towards to this way of thinking.

Would be fascinating to see how this all turns out in the next 2-3 years.

[selva2306]

Originally Posted by DigitalOne

There seems to be a clash between two school of thoughts.

The inherent ambiguity underneath OpenAI (what it was, what it is, and what it will be) is causing the drama. As you have rightly mentioned, there are two contrasting views on AI (as Science vs as Technology).

Hopefully, this whole drama leads to more decentralization of AI and open-source.

[DigitalOne]

Originally Posted by selva2306

The inherent ambiguity underneath OpenAI (what it was, what it is, and what it will be) is causing the drama.

Correct. From what I understand, OpenAI was founded as a non-profit, but later a "capped-profit" company was founded as a subsidiary to it. A "capped profit" company means that an investor cannot get more than x times (x =100 in case of OpenAI) of their investments as profit.

Microsoft may not back out of existing legal contracts with OpenAI but their pally relationship would end. Would be interesting to see where OpenAI goes from here.... Will Musk join forces back with OpenAI?

[DigitalOne]

Sam Altman to return as OpenAI CEO, new board members inducted

Link

[KrisTvpm]

Originally Posted by Fillmore

These days we see hoards of Advertisements on Youtube with different Indian startup training companies claiming to give you an edge by offering their courses on Prompt Engineering etc..
<snip>

This may be a too-late reply, yet FWIW I attended (only) the "master class" of such a startup, successfully wasted couple of hours of my lifetime.

The presenter chaps (IITians, per their claim) were straight into marketing, within 10min. of the session.

They showcased some 10 or so GPT-enabled tools to make ppts/diagrams/flowcharts etc which were somewhat useful, and then to their core-competency : marketing!

It was amusing to see their nice-cop/bad-cop show: one guy was technical, was giving some useful gyan, other was pure salesman & cringe at it!

They were pitching their so-called "bundle" of prompt engg/gen-AI/LLM/ChatGPT enabled learning-kit, starting with INR 1.5lakhs or so, bringing it down to INR 15K or so, over a painful-to-watch couple of hours of "show".

Overall, I learned I mustn't have wasted my 2hrs on this

[Samurai]

https://www.indiatoday.in/technology...529-2023-12-11

OpenAI confirms ChatGPT 4 has become lazy and that shows AI future may not come that early

[ValarMorghulis]

Originally Posted by Samurai

...
OpenAI confirms ChatGPT 4 has become lazy and that shows AI future may not come that early

This is disappointing.

But I wonder how the day will be like when AGI breaks through.
Will it be like "2003 a Space Odyssey" scenario?

[greyhound82]

Originally Posted by Samurai

OpenAI confirms ChatGPT 4 has become lazy and that shows AI future may not come that early

This autoregressive approach will work well for language as the prior distribution can be modelled using a finite number of tokens. This is why problems in computer vision are more difficult to solve because predicting the next state probabilistically is a bit complex due to its dimensionality. Current generative methods will fail for tasks involving hierarchical planning. There are researchers like Yan LeCun working on objective-driven AI for autonomous machines. A compact block diagram is given below (Image courtesy: Yan LeCun presentations)

[PGA]

I use ChatGPT to clarify a lot of doubts about agriculture and plant nutrition. It comes with very dated responses and often gets confused when discussing organic farming solutions. I was wondering if Iam using it in the manner it is meant to be. In that regard, I came across this
https://academy.dhruvrathee.com/chatgpt

I was interested in knowing if it will be of any help, any reviews / pointers of the course please.

[Fillmore]

Originally Posted by PGA

I came across this
https://academy.dhruvrathee.com/chatgpt

I was interested in knowing if it will be of any help, any reviews / pointers of the course please.

My curiosity got the better of me and i signed up for this.
It is all very generic stuff 85% of which is available on the net and other Y-Tube free videos. I was expecting some solid concrete stuff on Prompt Engineering but was disappointed. Not worth the money.

To protect from account sharing they made it such that the access login to the course is everytime via OTP sent to the registered mobile number, and you login via your Mobile number as the ID.

[sagarpadaki]

Originally Posted by Fillmore

My curiosity got the better of me and i signed up for this.
..........

I am not surprised considering that the course is curated by Dhruvrathee. You cannot expect quality value adding content from such mediocre content creators!

I pity those who follow him and such other similar content creators and fall for their trap of paid courses.

[SmartCat]

Originally Posted by PGA

I use ChatGPT to clarify a lot of doubts about agriculture and plant nutrition. It comes with very dated responses and often gets confused when discussing organic farming solutions. I was wondering if Iam using it in the manner it is meant to be.

The best way to get more out ChatGPT is to talk to it like you would to a person. Don't treat it like a 'search engine'.

If you think its response is outdated or confused, just say so to ChatGPT. Explain why you think so. Point out the mistakes in its answers. It will then take your feedback and try to give better answers immediately.

[novicefreak]

Try this

https://kissan.ai/

KissanAI

AI Agriculture Assistant for Bharat

https://kissan.ai/about

DISCLAIMER: The information and assistance provided by KissanAI are for general informational purposes only and are not intended to replace professional advice, consultation, or services. By using KissanAI, you agree to our Privacy Policy and Terms & Conditions. Please refer to these documents for more details on the use, limitations, and responsibilities associated with the Service.

[DigitalOne]

Originally Posted by SmartCat

The best way to get more out ChatGPT is to talk to it like you would to a person. Don't treat it like a 'search engine'.

If you think its response is outdated or confused, just say so to ChatGPT. Explain why you think so. Point out the mistakes in its answers. It will then take your feedback and try to give better answers immediately.

Correct. By sheer force of habit, I also include 'please', 'thank you' etc when I use ChatGPT .

On the other hand, for very specialized topics it may be better to refer the websites specializing in that domain. You may find the contact details of the specialists, Q&A sections, or premium/paid advice.