Commune

One day late, my apologies, but here are the links for the week: * TV soaps could be made by AI within three years. News from a parliamentary committee on art and culture around the accelerating speed of content creation in all industries. My guess is that this timeline is both right and wrong. Will it be possible to create a consistent, multi-part, ongoing entertainment as rich as (say) East Enders with AI in 3 years? It's very unlikely. Will significant parts of the process, from ideation to

Feb 24, 2024

Here are this week's links. * Cohere releases Aya: a model and dataset focused on support for 101 languages. Getting native language support in AI is a really important goal for equal access (and ultimately also for cultural preservation). Cohere's project crowdsourced and processed expert-level language data sets and translations. It's great to see that all the datasets are open source for others to build on top of. * Who makes money when AI reads the Internet for us? What happens if browser

Feb 16, 2024

One day late again, apologies! Travel has been heavy these past couple of weeks, so it's been tough to get ahead of the curve enough to post on time. It's been an interesting week with big shifts that could have long-term implications but muted commentary. A few things that stood out: * ChatGPT accelerates toward task automation. OpenAI is apparently working on a layer within ChatGPT that will allow the automation of complex tasks over backend services. I already commented on this task-centri

Feb 10, 2024

Here are this week's five links. If you're ready to have LLMs linked directly to your brain, then Elon Musk has something for you... * AI Tidbit's methods for securing LLMs from Jailbreaks: this is a useful concise overview, but more than anything, it serves to highlight how easy it is to get LLMs to do things their operators don't want them to. These powerful systems are extremely difficult to secure (one could argue impossible) since you effectively need an LLM even more powerful than the on

Feb 2, 2024

The week got away from me again, so we're one day late with these. Hopefully, they will still add something to your weekend: * Hugging Face and Google partner for open AI collaboration. Hugging Face is the biggest model-sharing community, and this extension focuses on making it easier to build and run more models. For Google, getting more workloads built on the infrastructure is an obvious win. This seems unlikely to be exclusive, so we'll probably see similar things happening on other cloud i

Jan 27, 2024

The end of the week turned into a big work crunch, so here we are with links on Saturday instead of Friday. Here's are this week's links with a distinctly science/technical lean in this case: * Google's AlphaGeometry can learn to solve geometric puzzles to the highest human standard. In a paper published in Nature, Google researchers show that designing a system that learns to solve complex geometry problems using a Neuro-symbolic approach is possible. I expect this will be one of the leading

Jan 20, 2024

Time for this for this week's links, which this week are a little bigger picture: * OpenAI introduces the ChatGPT Store: you can now put out (and use) popular fine tunes. It will be interesting to see to what extent the resulting selection of tools begins to look like existing app stores. One set of apps will be voice interfaces to existing apps (like Canva); others might be totally new. Microsoft also followed up by modifying its Phi-2 Open-Source SMB model license to the MIT license, making

Jan 12, 2024

Week 1 of 2024, and it's already Friday. In case you missed the update earlier this week, this newsletter is now coming to you via Ghost and not Substack. Happy New Year, and here are the links I found most interesting so far this week: * Tiny LLama: Putting a powerful LLM in your pocket. "Micro" is becoming a thing (which is excellent news). Tiny Llama is a 1.1B parameter model that was trained on fairly accessible hardware in just a couple of months (that's important) and is small enough to

Jan 5, 2024

I didn't expect to be writing another post over the new year, but here we are. This blog isn't intended to be political or socially activist (except perhaps insofar as AI impacts big societal decisions), but sometimes issues come to your door. Substack is a great technical platform and certainly lowered the barrier for me to start publishing again. Unfortunately, its leadership is choosing not to act against extremist violence-inducing content on its platform. The issues were highlighted in the

Jan 2, 2024