What does Claude think caused the FAA outage?
On November 15, 2026 8:22:23 PM CST, Steve Jones <[email protected]> wrote: >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
