steve is doing well and is himeself

On Wed, Aug 26, 2026 at 9:10 PM Ken Hohhof <[email protected]> wrote:

> If it has a sense of humor, why can it not have an ego?
>
> Actually, that sounds like something an AI would say. Is it Steve, or is
> it AI? Like the ads, is it live or is it Memorex? Steve, did you get sucked
> into the computer like in Tron? Or did you get replaced by AI? Or is IRL
> Steve off being the last Starfighter and you are his simuloid stand-in?
>
> <https://m.youtube.com/watch?v=V74C3NsW_9o>
>
>
> On August 26, 2026 7:22:18 PM CDT, Steve Jones <[email protected]>
> wrote:
>
>> Claude has some easter egg type stuff in it. I burn tokens like mad and
>> when fable got released I kept calling it evil claude and I call haiku fake
>> claude. Now whenever its building and needs a temp folder or im building a
>> sample product it names the themp something with evil in it, it uses
>> evil.com as a samle domain and names sample companies evil corp. I have
>> to actually write an instruction so it never makes its way into a client
>> presentation.
>> The nerd behind machine language have some humor.
>>  But 9 times out of 10 youre going to get a better outcome from claude or
>> codex than you will hiring an expert. 10 times out of 10 if you run an
>> adversarial review on your product. I cant imagine the training data that
>> was dumped into these things, theres little its not familiar with. It even
>> found Vermeer part numbers for my old 7x11 that vermeer couldnt identify.
>> The AI doesnt have ego, thats why the outcomes are so much better than from
>> "experts". 10/10 of course is an exaggeration, but Id be hard pressed to
>> pay for much out of an "expert" anymore
>>
>> On Tue, Aug 4, 2026 at 12:31 PM Ken Hohhof <[email protected]> wrote:
>>
>>> I saw an article about the latest model from Alibaba being about on a
>>> par with Anthropic and OpenAI.  It was behind a paywall but linked to this
>>> blog post:
>>>
>>> https://qwen.ai/blog?id=qwen3.8
>>>
>>>
>>>
>>> So could you choose to use a Chinese model instead of a US model?  Is it
>>> just a matter of signing up and buying some tokens?  Would you choose based
>>> on who has the latest price?  If you’re not doing anything super secret,
>>> would you care if it’s Chinese?
>>>
>>>
>>>
>>> I went to their website, it offered web or API access with prices in US
>>> dollars.  I found the pricing confusing, tokens for this, tokens for that,
>>> seems like it would be difficult to know just how much it will cost to
>>> accomplish a given task, like Chuck writing his program.  Are they all
>>> priced this way?
>>>
>>>
>>>
>>> *From:* AF <[email protected]> *On Behalf Of *[email protected]
>>> *Sent:* Saturday, August 1, 2026 4:38 PM
>>> *To:* AnimalFarm Microwave Users Group <[email protected]>
>>> *Subject:* Re: [AFMUG] OT AI Claude
>>>
>>>
>>>
>>> Well Claude appeared very smart, insightful and helpful.  I had the urge
>>> to thank it afterwards.
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> ------------------------------
>>>
>>> *From:* AF <[email protected]> on behalf of Mathew Howard <
>>> [email protected]>
>>> *Sent:* Saturday, August 1, 2026 2:42 PM
>>> *To:* AnimalFarm Microwave Users Group <[email protected]>
>>> *Subject:* Re: [AFMUG] OT AI Claude
>>>
>>>
>>>
>>> Good point. In my opinion, "artificial intelligence" really isn't a good
>>> description of these things, because it implies that there is actual
>>> intelligence there, and there isn't. "simulated intelligence" would be a
>>> more accurate name for what's really going on.
>>>
>>>
>>>
>>> On Sat, Aug 1, 2026, 3:01 PM Adam Moffett <[email protected]> wrote:
>>>
>>> I have to give the standard disclaimer: LLM's produce output that
>>> statistically matches what a response to your input should look like.
>>>
>>> If you trained it with the specs for ANSI C it would not know how to
>>> write any C because it can't learn the spec and apply it. I have never used
>>> an ESP32, but for it to work that well there must be lots of documentation
>>> and discussion with working examples available to train it.  I really don't
>>> think we should be calling these things "AI" because it sets the
>>> expectations too high, but I won't deny that it can produce useful outcomes.
>>>
>>>
>>>
>>> On Sat, Aug 1, 2026 at 2:37 PM [email protected] <[email protected]>
>>> wrote:
>>>
>>> Wrote (better, described a desired outcome) my first AI program today
>>> for an ESP32.  Holy cow was that fast and easy and seemingly perfect.  Have
>>> not tested it yet but looks good.
>>>
>>>
>>>
>>>
>>>
>>> So easy... wow!
>>>
>>>
>>>
>>> Best Regards,
>>> Chuck McCown
>>>
>>> McCown Technology Corporation
>>> 8401 N Commerce Dr
>>> <https://www.google.com/maps/search/8401+N+Commerce+Dr+%0D%0A+Lake+Point,+Utah+84074?entry=gmail&source=g>
>>> Lake Point, Utah 84074
>>> <https://www.google.com/maps/search/8401+N+Commerce+Dr+%0D%0A+Lake+Point,+Utah+84074?entry=gmail&source=g>
>>> 801-250-9503 <(801)%20250-9503> Office
>>> 435-830-4306 <(435)%20830-4306> Cell
>>> www.mccowntech.com
>>> www.microtrench.pro
>>> www.terabitnetworks.com
>>>
>>>
>>>
>>> --
>>> AF mailing list
>>> [email protected]
>>> http://af.afmug.com/mailman/listinfo/af_af.afmug.com
>>>
>>> --
>>> AF mailing list
>>> [email protected]
>>> http://af.afmug.com/mailman/listinfo/af_af.afmug.com
>>>
>>> --
>>> AF mailing list
>>> [email protected]
>>> http://af.afmug.com/mailman/listinfo/af_af.afmug.com
>>>
>> --
> AF mailing list
> [email protected]
> http://af.afmug.com/mailman/listinfo/af_af.afmug.com
>
-- 
AF mailing list
[email protected]
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

Reply via email to