#ai

waynerad@diasp.org

AI model comparison.

Compares input length, output length, input price (per 1 million tokens), output price (per 1 million tokens), and whether it supports vision.

Compares chat models, embedding models, image generation models, text completion models, audio transcription models, and speech generation models.

AI Model Comparison | countless.dev

#solidstatelife #ai #genai #llms

temp@sysad.org
ramnath@nerdpol.ch

https://youtube.com/watch?v=YHbNwygLlzs

Former #OpenAI #researcher and #whistleblower found #dead at age 26
Suchir Balaji, a former OpenAI researcher, was found dead in his San Francisco apartment in recent weeks, officials confirmed to CNBC.
Balaji left OpenAI earlier this year and voiced concerns publicly that the company had allegedly violated U.S. copyright laws in building its popular ChatGPT chatbot.
"The manner of death has been determined to be suicide," David Serrano Sewell, executive director of San Francisco's Office of the Chief Medical Examiner, told CNBC in an email
A 26-year-old former OpenAI researcher, Suchir Balaji, was found dead in his San Francisco apartment in recent weeks, CNBC has confirmed.

Balaji left OpenAI earlier this year and raised concerns publicly that the company had allegedly violated U.S. copyright law while developing its popular ChatGPT chatbot. "The manner of death has been determined to be suicide," David Serrano Sewell, executive director of San Francisco's Office of the Chief Medical Examiner, told CNBC in an email on Friday. He said Balaji's next of kin have been notified.

The San Francisco Police Department said in an e-mail that on the afternoon of Nov. 26, officers were called to an apartment on Buchanan Street to conduct a "wellbeing check." They found a deceased adult male, and discovered "no evidence of foul play" in their initial investigation, the department said.

News of Balaji's death was first reported by the San Jose Mercury News. A family member contacted by the paper requested privacy. In October, The New York Times published a story about Balaji's concerns.

"If you believe what I believe, you have to just leave the company," Balaji told the paper. He reportedly believed that ChatGPT and other chatbots like it would destroy the commercial viability of people and organizations who created the digital data and content now widely used to train #AI #systems. A spokesperson for OpenAI confirmed Balaji's death.

"We are devastated to learn of this incredibly sad #news today and our hearts go out to Suchir's loved ones during this difficult time," the spokesperson said in an email. OpenAI is currently involved in legal disputes with a number of publishers, authors and artists over alleged use of copyrighted material for AI training data. A lawsuit filed by news outlets last December seeks to hold OpenAI and principal backer Microsoft accountable for billions of dollars in damages.

"We actually don't need to train on their data," OpenAI CEO Sam Altman said at an event organized by Bloomberg in Davos earlier this year. "I think this is something that people don't understand. Any one particular training source, it doesn't move the needle for us that much."

waynerad@diasp.org

Low-level Guidance (llguidance) is a tool that can enforce arbitrary context-free grammar on the output of an LLM.

"Given a context-free grammar, a tokenizer, and a prefix of tokens, llguidance computes a token mask - a set of tokens from the tokenizer - that, when added to the current token prefix, can lead to a valid string in the language defined by the grammar. Mask computation takes approximately 1ms of single-core CPU time for a tokenizer with 100k tokens. While this timing depends on the exact grammar, it holds, for example, for grammars derived from JSON schemas. There is no significant startup cost."

"The library implements a context-free grammar parser using Earley's algorithm on top of a lexer based on derivatives of regular expressions. Mask computation is achieved by traversing the prefix tree (trie) of all possible tokens, leveraging highly optimized code."

guidance-ai / llguidance

#solidstatelife #ai #genai #llms

elijahu@pod.geraspora.de

Hat eigentlich schon jemand in aller seiner natürlichen Dummheit eine so genannte »künstliche Intelligenz« auf die bisherigen Ziehungen der Lottozahlen losgelassen, um eine »künstlich intelligente« Vorhersage der unüberschaubar multikausalen und damit für uns unvorhersehbaren Lottozahlen zu machen? Ganz ähnliche »Errungenschaften« sehe ich ja jeden Tag beim Nachrichtengenuss

#Blah #KI #AI

anonymiss@despora.de

American cops are using #AI to draft #police reports, and the #ACLU isn't happy

source: https://www.theregister.com/2024/12/12/aclu_ai_police_report/

"Police reports play a crucial role in our #justice #system," ACLU speech, privacy and #technology senior #policy analyst and #report author Jay Stanley wrote. "Concerns include the unreliability and biased nature of AI, evidentiary and memory issues when officers resort to this technology, and issues around #transparency.

#problem #news #future #dystopia #robocop #ethics #crime #court #prison #usa

varelsennormal@sysad.org

Der Erfolg der #AI wird auch in der #Leichtgläubigkeit der Menschen liegen!
Schon jetzt wird jeder technischen #Innovationen gehuldigt ohne sich für Fakten zu interessieren.
#Testberichte sind zu einer zaghaften Beurteilung verkommen!
Früher wurde AI als automatisierte #Prozesssteuerung bezeichnet!
Das trifft es meiner #Meinung nach noch immer am besten!
Wenn wir lernen unsere #Emotionen von den #Fakten zu trennen gibt es noch eine Chance!
Ansonsten gebe ich der #Menschheit noch 30 Jahre.
Dann ist #Schluss!

waynerad@diasp.org

Exa Websets purports to turn the whole internet into a searchable database.

"All AI startups building new LLMs chips that are post series A."

"All PhDs who have worked on developer products and graduated from a top university and have a blog."

"Obviously traditional search tools can't do these things. You don't even think to ask them that because they weren't built to be a database."

"So how do we do it? Well, we built the first web-scale embeddings-based search engine. Essentially, we trained an AI system to organize the whole web by meaning."

They claim "Exa's system knows when to use more compute to agentically research and verify each result. That means Exa Websets might take a long time to complete."

But it's not available now. You can join the waitlist. If this works as advertised, it'll be amazing.

Introducing Websets: A breakthrough toward perfect web search

#solidstatelife #ai #genai #embedding #searchengines

waynerad@diasp.org

The abject weirdness of AI ads. Not ads made using AI, ads made by AI companies about their AI products.

"I'm trying to find holiday gifts for my sisters. I open a bunch of tabs, I want my wife's advice."

"The company later pulled the ad after facing backlash for taking a sweet father-daughter exchange and automating it away."

"Many people pointed out that you could have just asked the stranger what type of dog they have, and maybe you would have found a friend alongside the dog's breed."

"An AI startup called Friend released a promotional video showing how lonely young people could have a virtual companion in the startup's AI device that they wear around their neck, instead of talking to others."

"Intelligence so big, you'd swear it was from Texas."

"Adapt your workforce at the speed of AI."

"AI that talks to cars and talks to wildlife."

The abject weirdness of AI ads

#solidstatelife #ai #advertising