In this panel, we will discuss how AI can enhance the design and implementation of embedded systems. Through this discussion, attendees will gain insights into the latest trends and developments in AI, as well as the challenges and opportunities for implementing AI in embedded systems and to assist in developing embedded systems. 

The panelists will share their experiences, research findings, and best practices, providing valuable insights for professionals, academics, and researchers alike. Overall, this panel promises to be a thought-provoking and informative discussion on the cutting-edge developments in the intersection of AI and embedded systems.

12:13:39	 From  Jim Norton : How cool would it be if some AI model, think GPT-4, could read through all of your project's datasheets and could put together a framework for a project and even write some device drivers for the developer!
12:14:56	 From  Nathan O. : Répondre à "How cool would it ..."

Finally we'd get what silicon vendor are supposed to provide us: drivers, proper example code and easy integration
12:15:03	 From  BobF : More clarity between the separate domains (use cases) requires some attention AI - ML - DL. Some thoughts amongst the panelists?                                 
n.b. AI is getting a lot of bad press which will impact upon the other domains which I think are less threatening .... I may be wrong here!
12:15:41	 From  Michael Lazarenko : Replying to "How cool would it be..."

Check embedd.it - that is exactly what we are working on!
12:15:59	 From  Jim Norton : Reacted to "Check embedd.it - th..." with 👍
12:16:20	 From  Gonzalo : Replying to "How cool would it be..."

It is very interesting and helpful, but what about industrial secrets, NDA, licencing exposing developers and company data?
12:17:00	 From  Nathan O. : A réagi à "Check embedd.it - ..." avec 😮
12:20:08	 From  Jim Norton : Replying to "How cool would it be..."

@Nathan O. 
I think CMSIS has gone a long way to meeting those goals.
12:21:30	 From  Jim Norton : Replying to "How cool would it be..."

Assuming ARM development...
12:21:55	 From  Nathan O. : Répondre à "How cool would it ..."

@Jim For RTOS and Core stuff yes. For actual peripheral I very rarely see drivers using the CMSIS API and personally I haven't used it much.
12:22:56	 From  Maxim Gorinov : Reacted to "@Jim For RTOS and Co..." with 👍
12:23:46	 From  Vishwa Perera : AI applications will definitely bring much ease to the programmer and product development. However, just to play devil's advocate what would be the negative impacts that AI applications bring to the embedded world?
12:27:11	 From  Steve Wheeler : I recently read an article (aimed at artists, mainly) that said that the terms of use for ChatGPT and other AI tools required you to indemnify them against legal action based on what you generated with them. Doesn’t this imply that any time saved in generating systems will need to be made up by more comprehensive study and testing of the generated code?
12:33:02	 From  Jacob Beningo : Reacted to "Finally we'd get wha..." with 👍
12:34:03	 From  Jim Norton : Replying to "How cool would it be..."

@Nathan O. 
NXP provides SDK's that actually use CMSIS for peripheral drivers such as ethernet, i2c, spi and uart.
12:36:52	 From  aniket : Replying to "How cool would it be..."

@Michael Lazarenko, Does your solution address ambiguity in data sheets? Some chip vendors have datasheets that are really poor - maybe even misleading
12:37:30	 From  Jack Albrecht : @Michael Lazarenko, Does your solution address ambiguity in data sheets? Some chip vendors have datasheets that are really poor - maybe even misleading
Furthermore, datasheets in different languages.
12:38:00	 From  Jim Norton : Reacted to "@Michael Lazarenko, ..." with 👏
12:40:35	 From  Kevin H : Philosophical question for the panel: Do you think society's increasing dependence on AI will adversely affect human intellect, curiosity, motivation, etc. over the long term?
12:45:04	 From  Gonzalo : Andre comments made me think that AI will have to be treated similarly to people (little scary) because what is fed with, what can share and answer will have to be under policies as we are working in companies. Above all two group will still existe: communities that share data and communities or companies that do not
12:46:11	 From  Jim Norton : How about AI generating "unit tests" to validate the accuracy of datasheets.
12:49:01	 From  Gonzalo : Replying to "Philosophical questi..."

I think it will affect as global connectiona and social media is already doing it. Big issues that needs to be faced sooner tan latter
12:49:44	 From  Steve Wheeler : I think there’s a misapprehension of my question. The article stated that if anyone sued based on the generated item, the terms required you to cover the AI company’s legal costs.
12:50:55	 From  Maxim Gorinov : Replying to "How cool would it be..."

To address this we have an Intermediate representation which is a structured extraction of the info relevant to driver development which user can adjust, edit or correct prior to driver generation.
12:53:24	 From  Sree Harsha Angara : Replying to "I think there’s a mi..."

Wow I would be very surprised if that actually finds any legal legs.
12:57:13	 From  Jim Norton : As far as becoming lazy because AI can do more for us, maybe what will happen is will be freed up to thinking about/work on bigger picture issues.

So perhaps we don't get lazy, but we just invent/think about things we didn't time for in the past.
12:57:18	 From  Gonzalo : I agree with Pete. I guess non-specialist person will be able to some things on embedded projects what could be good, but whenever real detailed issues appears and non-specialist could not face that the complexity will be revealed maybe too late in schedule for the developer team
13:00:41	 From  BobF : An answer for everything (ie. ChatGPT et al) ... doubtful !?
13:02:42	 From  Jim Norton : I question whether GPT is "intelligent".
13:02:59	 From  Jacob Beningo : Replying to "I question whether G..."

I would say no, but that’s me
13:03:14	 From  Jim Norton : Reacted to "I would say no, but ..." with 👍
13:03:21	 From  Michael Lazarenko : Reacted to "I would say no, but ..." with 👍
13:04:02	 From  Sree Harsha Angara : Replying to "I question whether G..."

This is a great primer https://www.youtube.com/watch?v=qbIk7-JPB2c
13:05:12	 From  Gonzalo : It was a good discussion, thanks to all!