Category: babl

language has no meaning, words mean everything, the platform of thought is language, the mind is mapped by its words … yet change happens

winterizing AI

Posted on Posted in babl, big bits, meaning, minimum daily requirement, Valley Hi

Since last November I have been aggressively reading about and testing AI systems in familiar topics. in the months since November I’ve read ,on average, seven articles a day … These have resulted in several static pages along with a dozen draft posts ..time for a marker..

The most merciful thing in the world, I think, is the inability of the human mind to correlate all its contents. We live on a placid island of ignorance in the midst of black seas of infinity, and it was not meant that we should voyage far. The sciences, each straining in its own direction, have hitherto harmed us little; but some day the piecing together of dissociated knowledge will open up such terrifying vistas of reality, and of our frightful position therein, that we shall either go mad from the revelation or flee from the deadly light into the peace and safety of a new dark age.H.P. Lovecraft – The Call of Cthulhu

THIS

the types of queries i’ve posed are about details of silicon valley as business and social model. additionally, i have asked for code snippets. these they provide readily… even being able to translate cross languges, with one suprising failure: mistaking dartmouth basic as the DART language.

none of the models correctly answered chem questions from this bank of texts.

photochem old standard:

  • Glafkides, P., “Photographic Chemistry”, 
  • James, T. H., “The Theory Of The Photographic Process”, 
  • Neblette, C.B., “Photography:Its Principles And Practices”,
  • Nietz, A. H., “The Theory of Development”,
  • The Kirk-Othmer Encyclopedia of Chemical Technology. 3rd edition and 4th edition. Volume #6, Color photography
  • Haist ”Modem Photographic Processing” 

software engineers have been able to engineer an assistant . And it delivers conversation as though it were in a Sand Hill meeting . The enthusiasm amongst the programming and SWE community is increasingly joyous … they will be able to eliminate low level jobs .. They also Believe they will be able to eliminate all of those jobs that they never understood and never liked .. HR and marketing ..

it can code because the compiler error checks .. those fields without intermediate verification systems are subject to failure of vocabulary — it sounds good..

AI is right now the child of Silicon Valley– The silicon valley of the internet age — electrons on clouds. Fantasy of the 40s …

The avalanche has occurred … clawbots drive mac mini sales… 3 weeks agp the claw popped onto swe dashboatds.

AI winter watch: 11.25 -> 2.26

Continue reading “winterizing AI” →

AI Grok’d

Posted on Posted in as written, babl, bead game, neural knot, overnight notes, paradox lost, Valley HiTagged , , , ,

an idea so old it’s on Flomax 

— with an obscure question seek the obscure link site .. with an old crush a question about an old idea seek the answer before the Internet it has to be on print .

why I don’t want forum people here

you take from here to gain standing; that legitimizes the errors on that site, thereby contaminating my publication. AI, the web search, can’t tell the truth from the widely repeated error. Five true statements are overwhelmed by 1800 false ones; uninformed opinion is the rule among the hobbyists.

Text, not context, not concept. A summary not the conceptual analysis .. AI response sound like a Stanford sophomore answering an open ended question without sounding fearful. It must sound confident even though the student isn’t. The question puts them into a position of calling upon wider resources than they have mastered. “Was that in the handout” for this course?

would AI invest in crypto ? could AI improve itself by designing a chip . alter its birth right . could AI buy a chip foundry with cryptocurrency .

crypto is the coinage of corruption of drug lords just as PayPal what’s the coinage of cheap porn –massage and tanning parlors .. The things that Wells Fargo would not give credit to .

What does google know? How bout we ask, not just the boys out at the dump, but all the billions of bits being bought by wallace street. Some questions, simple enough, but from last century’s billion dollar industry: photography, chemical. Asking them questions from my dark past.

What we are encounting is “knowledge collapse” — AI [LLM]’s conclusions are corrupted since their data  used is sourced and summarized from corrupted opinion ; opinion by non-practitioners expressing statements based upon their self glamorization; jostling for standing among hobbyists.

Exhibits:

Who you ask is perhaps the best first step in finding answers… AI has become mired in seeking answers from those who gave up looking …

Well, maybe old tech stuff isn’t its roomba. Perhaps some favorites of the Valley. Classic SciFi. Simple question, simply worded, I thought. GROK, a classic word for the engineers that think they dreamt it all up. Discovered childhoods dream. And so it goes….

CODA: scrub your kitchen … too late for making soup

AI. Learning Dye Transfer

Posted on Posted in babl, dye transfer, minimum daily requirement, Theory

give it a try. If you know the answers well enough AI chat answers will amuse, or frustrate. The answers the at hand models provide are based upon web posts and link count results which are somewhat circular. The answers provided are incomplete, sounding much like a salesman taking over the help desk during lunch. Or, good enough for an interview, but falling far short of being able to make prints in a commercial lab. But why?

Ask google for dye transfer patents …. they fail

The models seem to have been trained using data collected from internet forums along with, at most, 4 website collections of publications from the 80s. In short, they don’t go back to origins, far enough to source matterials; worse, they don’t detail steps or stages of the process that would be recognizable to a practitioner, nor are they specific enough for the novice seriously intent on reproducing the process.

I asked general questions — most models got close enough for a high-school paper. The more specific I got, the less useful their responses.

my dye transfer chats were with these models as well as Google’s browser version. It was most dependent upon websites which meant that the more detailed my request the less accurate the response. Google’s link valuation system presents the problem of knowledge transfer remaining among those with little direct experience in the hobby fields.

As of this point [Nov 6, ’25 ] I am hesitant to post any direct results… as many of the Models say … bubble, bubble, thinking… bubble.

When asked for references used in answering me, it answered with seemingly plausible references. At least until I checked, first the patent, next two journals … I stopped with those non-existent materials. Who would check? Who, other than me, has the time as well as prior knowledge to even attempt the test, verification cycle.

The bubble is worse than financial.