Re: ChatGPT ???

There's absolutely value to be extracted from these tools, but it's not
consistent as it can vary wildly. The challenge is how to extract the
upside and avoid the downside (or at least manage it). There's no doubt in
my mind that they can be useful, but you have to pick your battles
carefully at the moment.

Stuff like Claude code gets interesting when you set it up with automated
testing, but the same caveat applies here.

Regards,

Eben Bruyns


On Tue, 30 Sept 2025 at 12:38, Laughing Water <[email protected]> wrote:

> My take on it is that LLMs are English majors. I’ve gotten some really
> helpful programming tips from ChatGPT and Perplexity, but their code had a
> lot of commands that sounded right but didn’t exist in the chosen language.
>
> Best wishes,
> LW
>
> On Sep 29, 2025, at 4:54 PM, Eben Bruyns (as eben at sdk dot co dot nz) <
> [email protected]> wrote:
>
> Honestly, I've had that exact same thought! remember that intelligence <>
> knowledge. AGI ASI etc is completely different, I've seen what I consider
> to be sparks of a there "there". It behaves more like humans than I care to
> accept!
>
> I'm not sure if the current tech will get there, but if you cannot tell
> the difference does it matter?
>
> Regards,
>
> Eben Bruyns
>
>
> On Tue, 30 Sept 2025 at 11:46, Bradford Miller <
> [email protected]> wrote:
>
>>
>>
>> On Sep 29, 2025, at 4:19 PM, Eben Bruyns (as eben at sdk dot co dot nz) <
>> [email protected]> wrote:
>>
>> It's honestly uncanny how much it behaves like an over confident bad
>> junior dev!
>>
>>
>> So a successful AGI! Human-like intelligence doesn’t require actual
>> intelligence, just overconfidence and laziness! We’ve nailed it! Time to
>> take the rest of our lives off…
>>
>> Besides Mike Gazzaniga’s work on split brains (showing we do things and
>> then make up a story about why we did them; we never admit we don’t know
>> something), I’m reminded of so much 50’s SF it almost seems prescient.
>>
>
>