Current Affairs

Current Affairs

OpenAI and Axel Springer agree to show news in ChatGPT

Openai and axel springer announced an agreement to show summaries and links to news content in chatgpt. The source bounds the event: the release does not publish the full contract, financial terms or a general license for all news content. The story teaches how to separate access, license, attribution, link, training, product and undisclosed terms.

4 min 14 Dec. 2023

Current Affairs

Phi-2 shows what a small, bounded model can do

Microsoft research introduced phi-2, a small research model trained on curated and synthetic data. The source bounds the event: selected benchmarks do not make a small model a universal replacement, and the release carried research conditions. The story teaches how to compare parameters, data, task, benchmark, license, memory and inference cost.

3 min 13 Dec. 2023

Current Affairs

Gemini Pro reaches Vertex AI in preview: model, platform and product are not the same

Google opened Gemini Pro and Pro Vision in Vertex AI on December 13, 2023, still in preview. Separating model, platform and product reveals what was actually available and what remained a promise.

3 min 13 Dec. 2023

Current Affairs

Mistral AI confirms a €385 million funding round

Contemporary reporting documented mistral ai's confirmation of a €385 million funding round. The source bounds the event: the report does not publish all terms, and an attributed valuation is not cash, revenue or profitability. The story teaches how to separate capital raised, currency, valuation, rights, disbursement and technical capability.

3 min 13 Dec. 2023

Current Affairs

Microsoft and AFL-CIO agree on dialogue about AI and work

Microsoft and the afl-cio announced a partnership on artificial intelligence and labor. The source bounds the event: principles on information, training and dialogue did not guarantee jobs, collective bargaining or labor outcomes. The story teaches how to turn a labor partnership into a commitment, owner, timeline, right and verifiable outcome.

5 min 12 Dec. 2023

Current Affairs

Mixtral 8x7B: why activating fewer parameters does not mean loading a small model

Mixtral activates 12.9 billion of 46.7 billion parameters per token. A guide to separating compute, memory, benchmarks, variants and real openness.

4 min 11 Dec. 2023

Current Affairs

AI and jobs: task exposure is not job loss

The ilo published a global analysis of task and job exposure to generative artificial intelligence. The source bounds the event: exposure means neither a job created nor one destroyed, and outcomes depend on adoption, task, country and policy. The story teaches how to preserve occupation, task, scenario, country, denominator and mechanism when reading employment.

3 min 7 Dec. 2023

Current Affairs

Google introduces Gemini 1.0: how to read a benchmark before believing the headline

Google introduced Gemini 1.0 on December 6, 2023 as an Ultra, Pro and Nano family. Its results mattered, but variant, method, availability and limitations determine what a benchmark number can support.

3 min 6 Dec. 2023

Current Affairs

Gemini 1.0: how to read a benchmark without confusing model and product

Gemini Ultra scored 90.04% on MMLU with a protocol using up to 32 reasoning samples. A guide to comparing variants, inference, availability and the actual product.

4 min 6 Dec. 2023

Current Affairs

Wombo Me generates avatars: what happens to the photo

Wombo announced wombo me as an app for generating avatars from a photograph. The source bounds the event: the announcement does not establish fidelity, consent, retention or safety, and the store listing can change. The story teaches how to check submitted photo, permission, third parties, retention, deletion, cost and image use.

3 min 3 Dec. 2023

Current Affairs

AWS introduces new chips and Amazon Q at re:Invent

Aws announced the graviton4 and trainium2 chips during re:invent. The source bounds the event: the improvements were vendor comparisons under specific workloads and generations, not a universal advantage for every model. The story teaches how to compare chip, precision, memory, network, software, availability, power and total cost.

2 min 28 Nov. 2023

Current Affairs

Pika Labs Launches AI Video Platform, Raises $55 Million

Startup Pika Labs unveils Pika 1.0, a tool for creating and editing video from text and images, and announces a funding round led by Lightspeed Venture Partners.

4 min 28 Nov. 2023