Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > sci.physics > #895723 > unrolled thread
| Started by | Mild Shock <janburse@fastmail.fm> |
|---|---|
| First post | 2026-04-16 22:00 +0200 |
| Last post | 2026-06-09 20:10 +0800 |
| Articles | 3 — 1 participant |
Back to article view | Back to sci.physics
This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by
below is the oldest one visible, not the original post.
Univocity of AI: Cat Face Recognition (Was: Best way to use LLMs to augment academic research) Mild Shock <janburse@fastmail.fm> - 2026-04-16 22:00 +0200
Maybe buying AI Laptops is now amortisized? (Re: Univocity of AI: Cat Face Recognition) Mild Shock <janburse@fastmail.fm> - 2026-04-16 22:40 +0200
In 2028 BYD sells B100 performant laptop? (Was: Univocity of AI: Cat Face Recognition) Mild Shock <janburse@fastmail.fm> - 2026-06-09 20:10 +0800
| From | Mild Shock <janburse@fastmail.fm> |
|---|---|
| Date | 2026-04-16 22:00 +0200 |
| Subject | Univocity of AI: Cat Face Recognition (Was: Best way to use LLMs to augment academic research) |
| Message-ID | <10rrf4i$f12n$3@solani.org> |
Hi, I did the same using multiple LLMs in the past few weeks. Until ChatGPT degraded, they phased out the old models, and its now only 5.x. You get the effect of 4 eyes see more than 2 eyes. Now its for ChatGPT 5.x. kind of 1 eye and 1 eye- patch, plus completely brain amputated. Bye P.S.: Maybe the best AI application is this here: Does your cat bring home “gifts” too? https://zeromouse.com/ olcott schrieb: > On 4/16/2026 12:17 PM, Ross Finlayson wrote: >> On 04/16/2026 08:20 AM, olcott wrote: >>> (1) Progressively make the initial prompt more >>> unequivocal and succinct across five different LLMs. >>> >>> I use ChatGPT 5.3, Claude AI Sonnet 4.6 Extended, >>> Grok Expert, Gemini Pro, Copilot Think deeper >>> and occasionally NotebookLM for Deep Research >>> and deep analysis of specific documents. >>> >>> (2) Once initial prompt is unequivocal and succinct >>> across five different LLMs then test for consensus. >>> >>> (3) Once consensus is achieved carefully examine >>> actual verbiage of key source documents. For >>> academic research this involves direct quotes from >>> foundational peer reviewed papers. >>> >>> >> >> Maybe you should figure more how it's "univocal" than "unequivocal". >> > > by "unequivocal" I only mean that every LLM takes the > prompt to mean exactly the same thing after as many > as hundreds and hundreds of progressive refinements. > > Then after the prompt has been further refined to achieve > a complete consensus across all five LLMs this is a good > ballpark estimate of literally unequivocal. > > The final test is against foundational peer reviewed > research written by the well established leaders in > the field. > >> For example, you can give it an account of what "equality", >> according to Quine according to Russell, "is", and show >> that now it's removed and quite capricious and not very arbitrary. >> >> I.e., that's readily "equivocated". >> >> >> The philo-sophy needs an account of the philo-casuy, or as >> with regards to distinguishing and disambiguationg >> the "sophistry" and the "casuistry". >> > > Ultimately my system uses GUIDs for each unique sense > meaning of every word. > >> Or, anybody else's opinion is just as good, and not bad. >> >> So, "univocity" is a usual account against "the synthetic fragmentation >> into pluralistic accounts of wholes". that's been around forever, >> and is part of the philosophical canon. >> >> > >
[toc] | [next] | [standalone]
| From | Mild Shock <janburse@fastmail.fm> |
|---|---|
| Date | 2026-04-16 22:40 +0200 |
| Subject | Maybe buying AI Laptops is now amortisized? (Re: Univocity of AI: Cat Face Recognition) |
| Message-ID | <10rrhgn$f2hj$2@solani.org> |
| In reply to | #895723 |
Hi, Rumours are that the real winner is currently Google, advancing the art of LLMs and LRMs, while OpenAI and Anthropic only want to go IP, and have raining money. My brave AI Laptops can do the following: Q: Did Ramanujan consider this facny diophantine equations: x + sqrt(y) = 7 sqrt(x) + y = 11 A: The "Ramanujan Style" Longer answer generated locally, with y=9,x=4. What model did I use: https://lmstudio.ai/models/google/gemma-4-26b-a4b Performance of the AI Laptops: /* AMD Ryzen AI 7 350 with Radeon 860M */ 14 Tokens/sec /* Intel Core Ultra 7 258V with Intel Arc 140V */ 14 Tokens/sec Still a little lame. Maybe this explains why I don't use local models more often. But its a start! Bye Mild Shock schrieb: > Hi, > > I did the same using multiple LLMs in the past > few weeks. Until ChatGPT degraded, they phased > out the old models, and its now only 5.x. > > You get the effect of 4 eyes see more than 2 eyes. > Now its for ChatGPT 5.x. kind of 1 eye and 1 eye- > patch, plus completely brain amputated. > > Bye > > P.S.: Maybe the best AI application is this here: > > Does your cat bring home “gifts” too? > https://zeromouse.com/ > > olcott schrieb: >> On 4/16/2026 12:17 PM, Ross Finlayson wrote: >>> On 04/16/2026 08:20 AM, olcott wrote: >>>> (1) Progressively make the initial prompt more >>>> unequivocal and succinct across five different LLMs. >>>> >>>> I use ChatGPT 5.3, Claude AI Sonnet 4.6 Extended, >>>> Grok Expert, Gemini Pro, Copilot Think deeper >>>> and occasionally NotebookLM for Deep Research >>>> and deep analysis of specific documents. >>>> >>>> (2) Once initial prompt is unequivocal and succinct >>>> across five different LLMs then test for consensus. >>>> >>>> (3) Once consensus is achieved carefully examine >>>> actual verbiage of key source documents. For >>>> academic research this involves direct quotes from >>>> foundational peer reviewed papers. >>>> >>>> >>> >>> Maybe you should figure more how it's "univocal" than "unequivocal". >>> >> >> by "unequivocal" I only mean that every LLM takes the >> prompt to mean exactly the same thing after as many >> as hundreds and hundreds of progressive refinements. >> >> Then after the prompt has been further refined to achieve >> a complete consensus across all five LLMs this is a good >> ballpark estimate of literally unequivocal. >> >> The final test is against foundational peer reviewed >> research written by the well established leaders in >> the field. >> >>> For example, you can give it an account of what "equality", >>> according to Quine according to Russell, "is", and show >>> that now it's removed and quite capricious and not very arbitrary. >>> >>> I.e., that's readily "equivocated". >>> >>> >>> The philo-sophy needs an account of the philo-casuy, or as >>> with regards to distinguishing and disambiguationg >>> the "sophistry" and the "casuistry". >>> >> >> Ultimately my system uses GUIDs for each unique sense >> meaning of every word. >> >>> Or, anybody else's opinion is just as good, and not bad. >>> >>> So, "univocity" is a usual account against "the synthetic fragmentation >>> into pluralistic accounts of wholes". that's been around forever, >>> and is part of the philosophical canon. >>> >>> >> >> >
[toc] | [prev] | [next] | [standalone]
| From | Mild Shock <janburse@fastmail.fm> |
|---|---|
| Date | 2026-06-09 20:10 +0800 |
| Subject | In 2028 BYD sells B100 performant laptop? (Was: Univocity of AI: Cat Face Recognition) |
| Message-ID | <1108vrn$i9oc$1@solani.org> |
| In reply to | #895723 |
Hi, AI Laptops are technically evolving so much, the operating systems cannot evolve fast enough. Its like the cars in Bali geting bigger and bigger, but the streets are still narrow or bad. Quite a stunner was the RTX sparc announcement May 2026 that might put 1 PETA OPS AI Inferencing, or will it be machine learning, into your hands for $2000 ? There are a lot of things going on, like ARM SME2 unlocking even Apple Silicon for everybody, and Nvidia countering Apple Silicon with its own unified memory convential memory. What will be the future? I guess we might see Laptops having high bandwidth memory: TSMC’s CoWoS (Chip-on-Wafer-on-Substrate) is an advanced 2.5D packaging technology that vertically and horizontally integrates multiple distinct silicon dies https://www.nomadsemi.com/p/tsmcs-cowos-capacity But did you know that BYD started as a battery company and has now a semincoductor branch, already operating a 4nm? BYD electronics spinning off a laptop? MidAmerican Energy Holdings, a subsidiary of Warren Buffett's Berkshire Hathaway Inc, invested about US$230 million for a 9.89% share of BYD at HK$8 per share. https://www.nomadsemi.com/p/byd-semiconductor-deep-dive Bye Mild Shock schrieb: > Hi, > > I did the same using multiple LLMs in the past > few weeks. Until ChatGPT degraded, they phased > out the old models, and its now only 5.x. > > You get the effect of 4 eyes see more than 2 eyes. > Now its for ChatGPT 5.x. kind of 1 eye and 1 eye- > patch, plus completely brain amputated. > > Bye > > P.S.: Maybe the best AI application is this here: > > Does your cat bring home “gifts” too? > https://zeromouse.com/ > > olcott schrieb: >> On 4/16/2026 12:17 PM, Ross Finlayson wrote: >>> On 04/16/2026 08:20 AM, olcott wrote: >>>> (1) Progressively make the initial prompt more >>>> unequivocal and succinct across five different LLMs. >>>> >>>> I use ChatGPT 5.3, Claude AI Sonnet 4.6 Extended, >>>> Grok Expert, Gemini Pro, Copilot Think deeper >>>> and occasionally NotebookLM for Deep Research >>>> and deep analysis of specific documents. >>>> >>>> (2) Once initial prompt is unequivocal and succinct >>>> across five different LLMs then test for consensus. >>>> >>>> (3) Once consensus is achieved carefully examine >>>> actual verbiage of key source documents. For >>>> academic research this involves direct quotes from >>>> foundational peer reviewed papers. >>>> >>>> >>> >>> Maybe you should figure more how it's "univocal" than "unequivocal". >>> >> >> by "unequivocal" I only mean that every LLM takes the >> prompt to mean exactly the same thing after as many >> as hundreds and hundreds of progressive refinements. >> >> Then after the prompt has been further refined to achieve >> a complete consensus across all five LLMs this is a good >> ballpark estimate of literally unequivocal. >> >> The final test is against foundational peer reviewed >> research written by the well established leaders in >> the field. >> >>> For example, you can give it an account of what "equality", >>> according to Quine according to Russell, "is", and show >>> that now it's removed and quite capricious and not very arbitrary. >>> >>> I.e., that's readily "equivocated". >>> >>> >>> The philo-sophy needs an account of the philo-casuy, or as >>> with regards to distinguishing and disambiguationg >>> the "sophistry" and the "casuistry". >>> >> >> Ultimately my system uses GUIDs for each unique sense >> meaning of every word. >> >>> Or, anybody else's opinion is just as good, and not bad. >>> >>> So, "univocity" is a usual account against "the synthetic fragmentation >>> into pluralistic accounts of wholes". that's been around forever, >>> and is part of the philosophical canon. >>> >>> >> >> >
[toc] | [prev] | [standalone]
Back to top | Article view | sci.physics
csiph-web