BRN Discussion Ongoing

[Fullmoonfever]

Was googling some search words as I do and this person's CV came up. No longer with BRN but was, back in mid 2023, for 3 mths as a ML Intern working on MB EQXX.

What I found interesting is that it appears MB maybe weren't just testing or playing with Akida for "Hey Mercedes" KWS but also driver distraction via sensors if I read correctly?

AND...over a year after the EQXX demo in early 2022 if this person's dates are correct.

Nice to see pretty damn good anomaly detection performace against a high end, though slightly older, GTX 1080 GPU.

HERE

 

[DingoBorat]

Hi Smoothsailing 18.
Im just not completely trusting of Chat GPT or others, though obviously very good tools in the right context. See the Renesas announcement from december 2020.
First Akida IP License Agreement
cheers

With those kind of "tools" when you ask a question, the way the answer is given, depends a lot on how the question is asked, especially when you are seeking "its" opinion.
It seems to me, they are programmed to be a bit like a dog, that is eager to please and be pat on the head.

If you ask something like "what is the likelihood" it "knows" you are seeking confirmation of something and will try to supply evidence to confirm what you want.

A more reliable way to use it, as I understand it, would be to ask the question in a negative manner, like "what is the argument against" such and such a thing and you will possibly get a more helpful response (if you look at the results in a positive way, if you know what I mean, or even better ask the question in both/multiple ways).

That's just the way I see it currently and don't give much weight to it's "opinion".
Although it can be useful and now it's getting better, I intend to use it more in the Future, I'm getting kind of sick of seeing ChatGPT etc analysis everywhere...