GPT is at tool used in computer linguistics since more than 10 years.
It was just a matter of time until some brainless nerds would use it for
KI...
GPT just analysis and classifies text >the texts you give GPT. So
its not KI its the condensed shit some people want to throw at you.
Bu
Indeed, it can. It comes up with fake information. But now it is heavily
moderated to not allow that.
Em seg., 10 de abr. de 2023 às 16:33, H L V escreveu:
> Can it dream?
> Harry
>
> On Mon, Apr 10, 2023 at 11:49 AM Alain Sepeda
> wrote:
>
>> There are works to allow LLM to discuss in order to
In reply to Jed Rothwell's message of Mon, 10 Apr 2023 09:33:48 -0400:
Hi,
[snip]
>I hope that an advanced AGI *will* have a concept of the real world, and it
>will know the difference. I do not think that the word "care" applies here,
>but if we tell it not to use a machine gun in the real world,
In reply to Alain Sepeda's message of Mon, 10 Apr 2023 17:48:38 +0200:
Hi,
[snip]
>The real difference is that today, AI are not the fruit of a Darwinian
>evolution, with struggle to survive, dominate, eat or be eaten, so it's
>less frightening than people or animals.
The way a neural network lea
Can it dream?
Harry
On Mon, Apr 10, 2023 at 11:49 AM Alain Sepeda
wrote:
> There are works to allow LLM to discuss in order to have reflection...
> I've seen reference to an architecture where two GPT instances talk to
> each other, with different roles, one as a searcher, the other as a
> criti
There are works to allow LLM to discuss in order to have reflection...
I've seen reference to an architecture where two GPT instances talk to each
other, with different roles, one as a searcher, the other as a critic...
Look at this article.
LLM may just be the building block of something bigger...
I wrote:
> Food is contaminated despite our best efforts to prevent that.
> Contamination is a complex process that we do not fully understand or
> control, although of course we know a lot about it. It seems to me that as
> AI becomes more capable it may become easier to understand, and more
> t
LLM do not have intrinsic short or modifiable long term memory. Both require supplemental systems - reprompting of recent history or expensive offline fine tuning or even more expensive retraining.I think it’s fair to say no AGI until those are designed in, particularly the ability to actually lear
Robin wrote:
As I said earlier, it may not make any difference whether an AI
> feels/thinks as we do, or just mimics the process.
That is certainly true.
As you pointed out, the AI has no concept of the real world, so it's not
> going to care whether it's shooting people up
> in a video game,
GPT4 can have unlimited memory, right? Just give it access to a query
engine. Max token context length (input PLUS output) is 32k in the latest
model. GPT3.5 is 4096.
https://openai.com/pricing
Importantly, GPT4 has built 'world models' as a side effect of its
training. And when it predicts t
The most recent versions of Stockfish, the best chess engines, combines
"brute force", the usual branching algorithm, with NN. ChatGTP 4.0 (which
is actually quite similar to 3.5) uses plugins to be smarter. For example,
it can evoke wolfram alpha if it needs to make calculations. This modular
appr
In reply to Jed Rothwell's message of Sat, 8 Apr 2023 20:04:46 -0400:
Hi,
As I said earlier, it may not make any difference whether an AI feels/thinks as
we do, or just mimics the process. The
outcome could be just as disastrous if it mimics committing murder, as it would
be if it had murder "i
I wrote:
> The methods used to program ChatGPT and light years away from anything
> like human cognition. As different as what bees do with their brains
> compared to what we do.
>
To take another example, the human brain can add 2 + 2 = 4. A computer ALU
can also do this, in binary arithmetic.
Example, I used chatgtp to come up with a theory explaining the origin of
eukaryotes. The part I enhanced was something that chatgtp came up with.
In the theory of the origin of eukaryotes, we have discussed how colonies
of prokaryotic cells started transporting vesicles by kinesin, which
crossed
Robin wrote:
> For example, if asked "Can you pour water into
> > a glass made of sugar?", ChatGPT might provide a grammatically correct
> but
> > nonsensical response, whereas a human with common sense would recognize
> > that a sugar glass would dissolve in water.
>
> so where did it
In reply to Boom's message of Sat, 8 Apr 2023 20:26:43 -0300:
Hi,
[snip]
>It has a very short memory. It's something like 30kb.
...so's mine nowadays. :(
>If the conversation
>gets a little bit longer, it starts forgetting stuff, though it more ore
>less keep track of the sense of the topic.
It has a very short memory. It's something like 30kb. If the conversation
gets a little bit longer, it starts forgetting stuff, though it more ore
less keep track of the sense of the topic.
Em sáb., 8 de abr. de 2023 às 19:50, Robin
escreveu:
> Hi,
>
> The point I have been trying to make is tha
Hi,
The point I have been trying to make is that if we program something to behave
like a human, it may end up doing exactly
that.
Cloud storage:-
Unsafe, Slow, Expensive
...pick any three.
Yes, but have you tried to jailbreak it, this was a condition I told you
about. This type of answer is done by a moderation bot.
Em sáb., 8 de abr. de 2023 às 15:40, Jed Rothwell
escreveu:
> Boom wrote:
>
>
>> For those who used it in the first few days, when bot moderation was not
>> installed
In reply to Jed Rothwell's message of Sat, 8 Apr 2023 14:40:08 -0400:
Hi,
[snip]
>ME: ChatGPT is not considered artificial general intelligence (AGI). What
>qualities of AGI are lacking in ChatGPT?
>
>ChatGPT: ChatGPT, as a language model, has a narrow focus on generating
>human-like text based on
Boom wrote:
> For those who used it in the first few days, when bot moderation was not
> installed properly, of right now, if it is jailbroken, GPT works just as
> well as a very smart human. With a few tweeks (like making it use math AI,
> wolfram alpha which surpassed humans decades ago, or NN
21 matches
Mail list logo