chatgpt

Back Open Paginator
12.09.2025 18:47
r (@r@bsky.brid.gy)

*Son la primera generación que va a experimentar con #ChatGPT en su #educaciónuniversitaria. Lo más probable es que dé igual, pero veremos* Jordi Pérez Colome @jordipc Periodista Reportero de la sección de tecnología del periódico El País Preocupado por las consecuencias sociales que provocan




Show Original Post

Report

12.09.2025 18:44
2025 (@2025@bolesblogs.com)

ChatGPT Invents a New Language

As a Boles.ai experiment, I asked three AI Bots — Claude.ai and Gemini and ChatGPT — to create a language and then I asked them if the language was actually created by them or not. Because of the length of the responses, I created three separate articles demonstrating the language invention capabilities of each Bot. You can decide which language Bot was most effective and inventive. I used the same prompts for all three attempts.

PROMPT: (to ChatGPT) Invent a new AI language that we can use to communicate more succinctly and interestingly. This language should be written, understood by both humans and emerging AL and it must save time and space and be perfectly clear and easy to learn.

///

You want a language that’s shorter than English, clearer than email, and friendly to both humans and machines. English will be offended, but it’ll survive. Here’s a purpose‑built micro‑language that’s fast to write, trivial to parse, and hard to misunderstand.

STAMP

Structured Terse Annotated Message Protocol
A compact, role‑labeled writing system for human–AI communication.

0) Design goals

1) Core idea in one breath

A STAMP clause is a bag of labeled slots. Order never matters. Every clause must contain one V: slot (the predicate). Most clauses also have A: (agent) and P: (patient). These labels are the classic semantic roles used across linguistics, so both people and parsers instantly know who did what to whom. (Wikipedia)

A:Alex V:send.pst P:draft R:Bo L:email T:2025-09-11T17:00

“Alex sent the draft to Bo by email at 17:00 on 2025‑09‑11.”

This “role-first” approach mirrors frame semantics: you fill the roles relevant to the action. (Wikipedia)

2) Alphabet of slots (the only dozen you must memorize)

SlotMeaningTypical valueA:Agent/doerA:Alex, A:teamV:Verb/predicate (with dot-features)V:meet.pst, V:ship.fut.pfvP:Patient/thing affectedP:report, P:serverR:Recipient/beneficiaryR:client, R:BoL:Location or channelL:NYC, L:video, L:/urlT:Time (ISO 8601)T:2025-09-11, T:2025-09-11T09:30ZD:DurationD:45m, D:2hQ:Quantity/measureQ:3, Q:4kgM:Manner/method/styleM:quiet, M:expediteC:Cause/reasonC:deadline, C:policyH:Condition (“if…”)H:approved, H:X (if not)K:Evidence/confidenceK:seen, K:reported, K:p=0.8

Optional extras when needed:

Value types

3) Verb features (dot‑suffixes, stackable)

Examples:

4) Operators and punctuation

Examples

! A:you V:send.imp P:invoice R:@Acme T:2025-09-15

Command: send the invoice to Acme by the 15th.

? A:we V:meet.fut L:video T:2025-09-12T09:00Z

Question: shall we meet then?

A:@Rae V:merge.pst P:branch=feature-x > A:@Rae V:deploy.pst L:staging

Sequence/cause.

A:@Noor V:Xapprove.pst P:budget C:overlimit

Noor did not approve because over limit.

5) Minimal syntax (ABNF‑ish)

CLAUSE   = [OP] SLOT *(SP SLOT)OP       = "!" / "?"                       ; imperative or questionSLOT     = KEY ":" VALUEKEY      = "A"/"V"/"P"/"R"/"L"/"T"/"D"/"Q"/"M"/"C"/"H"/"K"/"I"/"G"/"S"/"B"VALUE    = TOKEN *("." FEAT)FEAT     = "prs" / "pst" / "fut" / "prg" / "pfv" / "hab" / "pas" / "imp"TOKEN    = 1*(ALNUM / "-" / "_" / "/" / "@" )

Deterministic. No hidden rules. A parser is about ten lines long and doesn’t need to guess.

6) Reading and writing, fast

English → STAMP

A:@You V:review.imp P:draft T:today ; ! V:send.imp P:final T:17:00
A:ops V:restart.pst.pas P:server C:security
H:legal-approved ; ! V:publish.imp T:2025-09-15

STAMP → English

7) Quantities, units, time

8) Evidence, certainty, and source

Make confidence explicit when it matters.

Example:

A:@Tariq V:arrive.pst T:2025-09-11T08:55 K:seen

9) Error handling and repair

A:@June V:booked.pst P:UNK L:hotel~ P:Suite-210

10) Composition patterns

Parallel tasks

A:@I V:write.prg P:intro + V:collect.prg P:cites + V:edit.prg P:fig-3

Alternatives

A:@We V:meet.fut L:studio | L:video T:2025-09-13

Conditional

H:weather-clear ; A:@Crew V:shoot.fut L:roof

11) Domain snippets

Legal

A:plaintiff V:file.pst P:complaint L:SDNY T:2025-09-10A:court V:grant.pst P:motion-to-dismiss H:failure-to-state-claim

Medical

A:pt-@J.D. V:present.pst P:cough D:7d M:dryA:clinician V:order.pst P:cxr K:guidelineA:lab V:return.fut P:results T:2025-09-12

Publishing

A:@Editor V:request.pst P:rev-2 C:clarity! A:@Author V:revise.imp P:sec-3 M:tighten Q:~250w

12) Micro‑lexicon (seed)

Verbs: meet, call, send, receive, approve, deny, file, sign, publish, draft, review, revise, schedule, delay, restart, deploy, ship, buy, refund, confirm, estimate, invite
Nouns: draft, contract, invoice, claim, server, release, figure, table, budget, policy, protocol, meeting, note, email, site
Adverbs/manner: urgent, softly, quietly, publicly, privately, cautiously, expedite

You can use any stem you like; STAMP cares about the labels, not the specific lexeme. This echoes long‑standing insights from information theory: clarity improves when you encode the parts that carry the uncertainty and strip out noise. STAMP encodes the roles and time, which are the high‑value bits in everyday coordination. (Harvard Math People)

13) A one‑page “cheat sheet”

MANDATORY: one V:COMMON:    A: (who)  P: (what)  T: (when)  L: (where/channel)FEATURES:  .prs .pst .fut  .prg .pfv .hab  .pas  .impOPS:       ! imperative   ? question   + and   | or   > cause→effect   X negateTIME:      T:YYYY-MM-DD[Thh:mm[Z|±hh]]QUANT:     Q:number[unit]    D:duration (m/h/d)EVIDENCE:  K:seen|heard|reported|computed|inferred|p=#

14) Why this works (and why it’s not Lojban with a haircut)

15) Training plan and rollout timeline

You can make this operational in one week. Starting September 11, 2025:

16) House‑style macros (optional, not magic)

You can define project‑local shorthands that simply expand to STAMP. Keep them explicit.

MACRO: ack  →  V:confirm.prs P:receiptMACRO: asap →  M:expedite T:soonMACRO: ping →  V:notify.prs L:chat

Usage:

A:@You ack P:files ; A:@I ping R:@You T:2025-09-12T09:00

17) Full examples

Decision note

A:@Board V:approve.pst P:budget-2026 Q:12.4M K:recordedA:@CFO V:allocate.fut P:dept-R&D Q:4.1M > A:@R&D V:hire.fut Q:6

Complex instruction

! A:@You V:compile.imp P:refs-appendix L:/shared/dir+ V:check.imp P:citation-style M:strict+ H:missing-doi ; V:add.imp P:doi K:computedT:2025-09-13

Repair after mistake

A:@Ops V:deploy.pst P:build-127 L:prod~ A:@Ops V:rollback.pst P:build-126 C:regression

18) Interop: dead‑simple JSON mapping

STAMP clause:

A:@Mina V:send.pst P:contract R:@Bo L:email T:2025-09-12T10:00Z K:reported

JSON:

{  "A": "@Mina",  "V": {"lemma": "send", "tense": "pst"},  "P": "contract",  "R": "@Bo",  "L": "email",  "T": "2025-09-12T10:00Z",  "K": "reported"}

One line in, one object out. Every time.

19) Pedagogical minimalism

Teach four things and you’re fluent:

  1. A V P
  2. T L when time or channel matters
  3. Verb features .pst/.fut/.prg
  4. ! and ? for commands and questions

Everything else is sugar you add when you care about nuance.

20) Final sanity check

If you want this sharper still, we can add a tiny validator or a VSCode snippet pack to auto‑colorize slots. But you can start using STAMP immediately, which is the whole point.

References (context for the design)

Now take a breath, and then write your next message in STAMP. If English complains, tell it the future arrived on time.

///

PROMPT: Is this language totally new, unique and created only by you?

///

Short answer: the specific system described here as “STAMP: Structured Terse Annotated Message Protocol” — with its fixed slot labels (A: V: P: …), dot‑suffix verb features (.pst .pfv .imp), and operator set (! ? + | > X) — was coined and specified here today. In that sense, yes: it’s newly created in this exchange. The acronym, however, is gloriously non‑unique in the wider world, and the underlying design principles stand on the shoulders of prior work in linguistics and formal/agent languages. Novel, yes; ex nihilo, no. If you want to publish or trademark it, you’ll want to disambiguate the name from several unrelated “STAMPs.” (IETF Datatracker)

What’s original here

In short: the parts aren’t new, the arrangement is.

Name collisions you should know about

“STAMP” already names other things. None of these are the language you and I defined, but they exist and will confuse reviewers and search engines:

If uniqueness of the name matters, consider “STAMP‑AVP” or “STAMP‑Protocol (Human‑AI)” in formal venues. Otherwise, keep STAMP and add a subtitle the first time you introduce it.

Intellectual lineage (what this borrows from, on purpose)

If you feel seen by that list, good. Reusing the best ideas is how new tools avoid being precious and useless.

Timeline of the ideas that fed STAMP

Bottom line

If you want a belt‑and‑suspenders claim to novelty, publish a brief spec with examples and a permissive license, include a “prior work” section citing the items above, and you’ll be the first citation anyone can point to when they try to reinvent this in six months with different punctuation.

#ai #bot #chatgpt #communication #compute #english #grammar #invention #language #meaning #open #syntax





Show Original Post

Report

12.09.2025 18:38
StopDisinformation (@StopDisinformation@mastodon.social)

*Son la primera generación que va a experimentar con en su . Lo más probable es que dé igual, pero veremos*

Jordi Colome profesor en la universidad Uc3m

@jordipc

Periodista Reportero de la sección de tecnología del periódico El País
Preocupado por las consecuencias sociales que provocan




Show Original Post

Report

12.09.2025 18:30
RaphaelWimmer (@RaphaelWimmer@hci.social)

A group of master's students had a problem with serial communication between #ESP8266 modules. They asked #ChatGPT for help.

The wise LLM argued that serial communication can only run at 9600 bps (very wrong) which would be too slow for their use case. ChatGPT then suggested that the students should switch to I2C instead - which would have cost them a lot of time and would not have worked well for their use case.




Show Original Post

Report

12.09.2025 18:27
mattzehn (@mattzehn@indieweb.social)

Versteht KI die Sprache doch? – Antworten auf Ihre Einwände

#KI #LLM #ChatGPT #Philosophie #Wahrheit #Verstehen #Denken #Bewusstsein #Simulation

Lesen: matthiaszehnder.ch/wochenkomme
Hören: buzzsprout.com/1788913/episode
Schauen: youtu.be/Q4yWoAuxEH0





Show Original Post

Report

12.09.2025 18:11
mike (@mike@gts.minus1.com)

Content warning:Babies and LLMs

Last week, I spent some time with Haley (one year old) doing serious research into ‘putting things into other things’.

We’d brought her a soft toy squirrel and, inevitably, she had found the cardboard box it came in much more interesting. There followed a considerable period of putting various objects into the box and taking them out again. This was pretty interesting, but the discovery that closing the box made the items disappear was even more fascinating. Also, if you moved the box, all the things in it moved at the same time. Brilliant.

Next up was a small basket into which went a toy monkey. Taking the monkey out, Haley replaced it with her smaller monkey toy. However, even after several tries, it became obvious that both monkeys would not fit in the basket at the same time. So, containers had the property of capacity.

Upping the stakes somewhat, we moved onto a toy ark with cutout animal shapes. It had both windows and doors. Holding it on end, it didn’t work the same as the other containers, as when you put an animal it it, it didn’t stay in the ark, but fell out of the window on the other end. Some containers are more complicated than others.

Overall, we spent about an hour investigating the idea of containers, and as always with very small children, Haley’s ability to construct sophisticated models of how the world works impressed me hugely.

I also came away with some important ideas.

In the next year or two, Haley will create exponentially more complex models of the world, and she will do this without language.

If Haley was an LLM, I would have spent many hours telling her hundreds of stories about boxes and monkeys, with the result that she would have been able riff endless variations on container stories. But that doesn’t mean she would understand the underlying physical concepts that govern the interaction between objects. She’d just be a good story teller.

But Haley is human, and will have developed a huge repertoire of knowledge way before she gets to language, which, admittedly, is the icing on the cake.

But all LLM’s have is the icing and no cake. That’s not to say that what they can do is not useful or indeed remarkable, but without the benefit of 300 million years of evolution and the embodied experience that Haley has, I think they will remain excellent storytellers, albeit very useful ones.

#chatgpt #machinelearning #openai




Show Original Post

Report

12.09.2025 18:04
AAKL (@AAKL@infosec.exchange)

Yes.

"The subtle quirks that once defined regional or communal identity now risk being diluted, replaced by a kind of synthetic 'GPT English'.”

Computerworld: Is AI changing our language? computerworld.com/article/4055 @MikeElgan #AI

Results showed "a surge in GPT words in the 18 months after ChatGPT’s release. The words didn’t just appear in formal, scripted videos or podcast episodes; they were peppered into spontaneous conversation, too."

UCLA analysis, from July: ChatGPT Is Changing the Words We Use in Conversation scientificamerican.com/article #ChatGPT

Arhus University, Denmark, from April: Will AI shape the way we speak? The emerging sociolinguistic influence of synthetic voices arxiv.org/html/2504.10650v1




Show Original Post

Report

12.09.2025 18:01
tugatech (@tugatech@masto.pt)

Google desenvolve método híbrido para acelerar os chatbots de IA
🔗 tugatech.com.pt/t71757-google-

#cascata #chatbots #ChatGPT #google #ia #llms #programação #sem #tecnologia #velocidade 




Show Original Post

Report

12.09.2025 17:26
display (@display@friendica.quebec)

"L’intelligence artificielle n’existe pas, c’est la bêtise artificielle qui caractérise notre époque" déclare...

'L’intelligence artificielle n’existe pas, c’est la bêtise artificielle qui caractérise notre époque' déclare...#IA #OPENAI #ELONMUSK #CHATGPT
Finirons-nous par écrire, parler et penser comme ChatGPT ? - RTBF Actus




Show Original Post

Report

12.09.2025 16:32
aimainainu (@aimainainu@mastodon-japan.net)

Chat GPTペット擬人化で大爆笑！ 衝撃の結果に！！！ そして超ビッグな傘で安心（PR含） : WITH LATTICE Powered by ライブドアブログ
n-lattice.blog.jp/archives/442

#ChatGPT #ペット #ペット擬人化




Show Original Post

Report

12.09.2025 16:29
desdeabajo (@desdeabajo@mastodon.social)

📱| Estudios de MIT y OpenAI encontraron que tanto el comportamiento del modelo como el del usuario pueden influir en los resultados socio-emocionales.

desdeabajo.info/otras-noticias





Show Original Post

Report

12.09.2025 16:01
r (@r@bsky.brid.gy)

#OpenAI and Microsoft sign preliminary deal to revise partnership terms. Via @arstechnica #AI #ArtificialIntelligence #ChatGPT #LLM 💻 🤖 🧠

OpenAI and Microsoft sign prel...




Show Original Post

Report

1 ...1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 ...1647
UP
created by:  Chefblogger.me - Eric on Mastodon