- AI Time to Impact
- Posts
- . . AI: Altman v Musk jabs, AI's impact on knowledge workers, AI animation (11.10.23)
. . AI: Altman v Musk jabs, AI's impact on knowledge workers, AI animation (11.10.23)
Friends, for your Friday night, we’ve got some AI entertainment. There’s a Dad Joke Battle and some very cool AI animation demos.
As always, these are the stories the AI community is talking about most, with their Time to Impact estimated by an ensemble of LLMs. I use a lot of technology in creating these (today ChatGPT caught three typos in my writing!), but I am responsible for the editorial contents. :)
Now here’s the news.
Marshall Kirkpatrick, Editor
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/ec9f52a9-3537-4677-8438-21c95f8ca957/Intro_1__6_.png)
Altman and Musk Engage in Chatbot Battle With Mocking Tweets
First impacted: AI researchers, AI industry professionals
Time to impact: Whatever
OpenAI's CEO Sam Altman used his firm's new GPT tool, which allows users to make their own chatbot, to mock Elon Musk's new AI chatbot, posting the words "GPTs can save a lot of effort" with a photo on X of an OpenAI GPT being told to use "cringey boomer humor in a sort of awkward shock-to-get-laughs sort of way." In the screenshot the app said it would name itself Grok, which is the name of Musk's AI. Musk responded with what he said was Grok's response, roasting GPT-4 ("more like GPT-Snore!"), and including fresh examples of the aforementioned cringey boomer humor, as alleged, but without any apparent self-awareness. It was the most-talked-about topic among AI community members on X all day. [via @sama] Share by email
Studies: Generative AI Disrupts Online Freelancing Market
First impacted: Online freelancers, Less skilled workers
Time to impact: Short
John Burn-Murdoch, chief data reporter at the Financial Times, has posted an X thread about AI's impact on knowledge workers. He combines one study finding that the launch of ChatGPT led to a shift in the online freelancing market, resulting in fewer job opportunities and lower earnings, even for highly skilled freelancers, with another study from Harvard Business School that suggests that while ChatGPT can advance productivity and work quality, it also poses difficulties in tasks needing careful interpretation, with the most notable performance enhancements seen among less skilled workers. More than a million people have seen the thread. [via @jburnmurdoch] Share by email
Runway Announces Motion Brush for Gen-2 Software
First impacted: Graphic Designers, Digital Artists
Time to impact: Short
Runway's upcoming Gen-2 software is set to include the new Motion Brush, a tool that allows users to add controlled movement to their designs just by dragging a brush tool over an image. The demo video shows images of objects like vehicles, flames, and smoke animated by the brush. [via @runwayml] Share by email
Pika Labs to Launch AI-Created Animation Product
First impacted: Animators, 3D Animation Artists
Time to impact: Medium
Pika Labs is preparing to launch Pika 1.0, a new product offering text-to-video animation, as shown in a 3D teaser that highlights animated characters' wide array of movements and expressions. The demo video is made up of short clips and calls itslef "the future of animation." [via @pika_labs] Share by email
Zephyr Model Training Code Released for Public Use
First impacted: AI developers, Data scientists
Time to impact: Short
Zephyr model training code is now publicly available on Hugging Face. Zephyr was the result of an international open source team's recent work extending Mistral. [Robust recipes to align language models with human and AI preferences] Share by email
Google Cloud Launches Full-Stack Training Offering
First impacted: Machine Learning Engineers, Data Scientists
Time to impact: Short
Google Cloud has launched Cloud TPU Multislice Training, a new and user-friendly solution compatible with TPU v4 and v5e, used to carry out the biggest publicly disclosed LLM distributed training job on a network of 50,944 Cloud TPU v5e chips, reportedly exceeding the AI accelerators of the TOP1 Supercomputer Frontier at Oak Ridge National Laboratory. The company says this system, incorporating various elements of the JAX training stack, provides efficient compilation, robust management, scalability, and advances popular machine learning frameworks like JAX and PyTorch. [the world’s largest distributed LLM training job on TPU v5e] Share by email
That’s it! More AI news on Monday!