[kwlug-disc] ChatGTP + Bing: we're entering a new paradigm and it's mind-blowing

[Khalid Baheyeldin]

On Thu, Feb 16, 2023 at 2:04 PM Mikalai Birukou <mb at 3nsoft.com> wrote:

>
> > There is a huge gap between reality and expectation ...
>
> And we, humans in statistical general, hence skewed and dominated by
> normies, we have a brain that bridges that gap, or glosses it over. It
> is amazing to observe. This is as much about humans as it is about
> external tech.
>

Regardless of the nitty gritty technical details ...

We know that ChatGPT can pass tests with average scores. On TV there
were students in local universities that were using it as a legit assistant
(e.g. expand on what the assignment is), or shady (e.g. do their work
for them). The latter can be your future doctor, civil engineer, lawyer,
judge, ...etc.

But even if there is no cheating and the universities are able to curb it.

Imagine that AI gets adopted in certain domains, and this 70% score is
not taken into account, think of all the pitfalls: legal cases on trial with
wrong interpretation of the law, or a doctor that wrongly diagnoses an
illness based on what AI says ...

And so on ...

That is what I mean by the gap between reality and expectations.
If expectations are "it is AI, and therefore more accurate than what
humans can do", then there will be a lot of problems ...

If it is just a starting step in a process, then MAYBE it would be useful
...

Related ...

Here are several other cases of non-factual results

https://www.theverge.com/2023/2/15/23599072/microsoft-ai-bing-personality-conversations-spy-employees-webcams
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://kwlug.org/pipermail/kwlug-disc_kwlug.org/attachments/20230216/dc4c2523/attachment.htm>