[ih] History of AI and Internet

[John R. Levine]

> I know this is naive and idealistic, but what do we learn about models of languages from such exercises?
> Is this just engineering?  Or do we learn something about how the brain works?

Real answer: nobody knows.

Slightly longer answer: LLMs are a weighted function of billions, maybe 
trillions of weights.  It's so many that as far as I know there is no way 
to tell what the model really is other than poking at it from the outside.

People have built neural networks with threshold gates that operate sort 
of like biological neurons, but I don't see any reason to believe that 
the way they interconnect is like the way actual neurons do.

R's,
John

>> On Jun 23, 2026, at 13:34, John Levine via Internet-history <internet-history at elists.isoc.org> wrote:
>>
>> It appears that Brian E Carpenter via Internet-history <brian.e.carpenter at gmail.com> said:
>>> On 23-Jun-26 17:25, william yeager wrote:
>>>>
>>>>> On Jun 22, 2026, at 7:54 PM, Brian E Carpenter via Internet-history <internet-history at elists.isoc.org
>>> <mailto:internet-history at elists.isoc.org>> wrote:
>>>>>
>>>>> but expert systems were not real AI
>>>>
>>>> Rubbish ! (Steven Hawking’s favorite denial). They WERE real AI at the time they existed.
>>>
>>> Fair enough, but all the expert systems I saw were basically rules created by human experts.
>>
>> Well, yeah, that was how people did AI when I was in school in the 1970s. Try to
>> come up with more or less realistic models of the way people think.
>>
>> I think everyone was surprised when Google fed the parallel English and French
>> texts of the Canadian Hansard into a machine learning system, and got a much
>> better translator than any of the existing ones that were built with explicit
>> language models. Now most AI is variations of dump in a lot of data and let the
>> algorithms turn it into whatever model they get, but that's a huge change in
>> mindset from AI 50 years ago.