This week in the news: OpenAI follows Anthropic in keeping its newest security model restricted, LinkedIn gets into AI training labor market, and AI Robot demos keep getting more violent (someone rein in the marketing department!).
On to this week's top stories:
- Anthropic just launched Claude Design, an AI tool that turns prompts into prototypes and challenges Figma. Anthropic enters the fray for prompt-based app builders with Bolt, Lovable, Vercel, Figma, and others. There's a logic to this since much of the success of these other projects is using Anthropic Claude underneath. I can also see it being used to build micro-apps in the enterprise. On the other hand, it's hard to imagine they will really beat focused companies that try to support the whole "idea to app to deployment" lifecycle. Anthropic has so many fish to fly, one wonders whether or not this product will receive enough attention. My bet would be that this becomes a great enterprise internal app builder (and is valuable for that as that class of application grows), but doesn't really displace the growing ecosystem of other app-builders.
- Chinese firm slammed for using ex-employee’s data to create ‘AI human’ to continue working. A Chinese company created an avatar of a former worker with their consent and saw considerable backlash for this (and legal challenges). It seems that the company may have used personal communications for some of the training, which may be illegal, and it seems a risky thing to do to not only clone someone's function but also their likeness. One would at least expect a royalty cheque for use of a personal image. Having said all this, the same process (minus the avatar) is happening wholesale across thousands of large organizations. As employees are pushed to use more and more AI, they are also being asked to document their roles in text files: laying the groundwork for layoffs and replacement by AI. It may be economically rational, but training your replacement is a hard pill to swallow.
- The Spirit of a Writer in the Age of AI. Great little video by writer Robert Greene on writing in the age of AI. His key point is that you have to have the creative originality to steer the process. I think he's right, and he's also right that (right now) originality has to come from a human. Even if AI can create something weird and original, it takes a human to recognize the brilliant ideas and curate/shape those. Notice I say "right now" since I'm pretty certain that we will also see great "AI writers" that shape original work just as well or better than humans can.
- A team at Meta Open Sources AI4AnimationPy: Fully Python Native AI-Driven Animation. One of those things that seems small but could have a big impact. Creating animations with Neural Networks is hard, and testing them often involves using third-party tools like Unity to show & validate the results. This new open-source framework means animations can be created and visualized in Python with standard libraries. This will enable animation creation systems to iterate much faster.
- HY-World 2.0: A Multi-Modal World Model for Reconstructing, Generating, and Simulating 3D Worlds. At the same time, Tencent released a new model that allows the creation of 3D interactive worlds. The model creates maps, simulates some physics, and creates a relatively style-consistent world from a series of prompts. This still won't be good enough for high-fidelity game design, but we can't be far away from meaningful games being created using technology like this.
Wishing you a great weekend.
