Machines of Loving Disgrace

In partnership with

Hello readers,

Welcome to the AI For All newsletter! Another CEO of an AI company shares yet another rosy vision for the future, robotics is becoming dependent on LLM progress, and the Tesla robotaxi event is worse than we thought. Let’s dive in!

Machines of Loving Disgrace

Anthropic CEO Dario Amodei recently published a blog post titled Machines of Loving Grace about what a future with “powerful AI” might look like. It is as long as it is self-aggrandizing despite attempting to appear grounded. Unlike Sam Altman, Amodei is careful to avoid, as he puts it, “grandiosity,” “sci-fi baggage,” and the “perception of propaganda,” only to then go into exhausting detail about a future based entirely on an imaginary technology. What’s this? Anthropic is currently trying to raise money?

There are two parts of the blog post that I take the most issue with:

“The basic development of AI technology and many (not all) of its benefits seems inevitable (unless the risks derail everything) and is fundamentally driven by powerful market forces. On the other hand, the risks are not predetermined and our actions can greatly change their likelihood.”

Unless the risks derail everything? So, it’s not inevitable. Powerful market forces? What are those exactly? The unsustainable, unprofitable, and misguided faith of Big Tech? Wall Street vibes? Markets are brittle. As long as AI doesn’t outgrow LLMs, the bubble will burst, and AI risks another winter. What’s funny is that there’s no principled reason to think AGI is inevitable. We could just like, you know, not figure it out. The risks are not predetermined? That’s right, they’re already here, and they were all predictable: disinformation, copyright infringement, scams, social engineering, etc.

“I don’t know if this world is realistic, and even if it is, it will not be achieved without a huge amount of effort and struggle by many brave and dedicated people […] If all of this really does happen over 5 to 10 years—the defeat of most diseases, the growth in biological and cognitive freedom, the lifting of billions of people out of poverty to share in the new technologies, a renaissance of liberal democracy and human rights—I suspect everyone watching it will be surprised by the effect it has on them.”

The problem is hypothesizing about a non-existent technology as a pretense for further fundraising. Man, where are all the time travel startups? If you call what these people are trying to build what it actually is, the “everything software,” then it’s suddenly not so tantalizing. It just sounds doomed to fail. Anthropic expects to create a software program that makes scientific discoveries, solves poverty, and achieves world peace — as long as nothing goes wrong, regular people make most of the sacrifices, and you give them billions of dollars. Don’t you want a prosperous future!?

LLMs are woefully insufficient for robust, secure, and reliable software, much less AGI. Their ability to reason is severely limited and patching each new hack that’s found is as demoralizing as plugging holes in a sinking ship. Robotics is precariously hitching its wagon to LLMs. At this year’s World Robot Conference, numerous Chinese companies passed off animatronic puppets as advanced androids, with some expressing hope that advancements in ChatGPT will improve their robots. But wait, if LLMs can be hacked, can LLM-controlled robots be hacked? The answer is yes.

Speaking of robots, let’s revisit Elon Musk’s Cybercab event. I wrote in last week’s newsletter that Elon was trying to keep Tesla’s stock price inflated, and it turns out that didn’t work. This charade of an event featured Tesla bots that were remote controlled, vague timelines, no mention of the many regulatory hurdles and Tesla being far behind its competitors, and an inexplicable design for the robotaxi.

And then there’s the Robovan, Tesla’s “answer” to trains and buses, which carries a whopping 20 people. Only in America where public transportation is seen as being for the “lower classes” does an idea like this get any traction. Instead of the tech industry constantly trying and failing to reinvent the train, why not just build an actual train and bus network and supplant the asphalt hellscape that is American car infrastructure? Reminder: having lots of money ≠ having good ideas.

Elon Musk, Mark Zuckerberg, Sam Altman, these men should not be mythologized. Their many moral, intellectual, and business failings should not be swept under the rug by access journalists and the legions of envious frat bros on social media pretending to be savants. From tech zealots to athletes, society will ignore the faults of those it deems valuable, thinking them “tortured geniuses.” This toxic relationship catalyzes an atmosphere of zero accountability that the powerful exploit. Why is this normalized? Because many want to believe that everyone at the top deserves to be there, lest the functioning of the meritocracy be brought into question.

🔥 Rapid Fire

• NVIDIA releases Llama-3.1-Nemotron-70B-Instruct, surpassing GPT-4

• Mistral AI releases Ministral 3B and 8B models for on-device and edge

• Apple researchers publish paper showing limitations of reasoning in LLMs

• Google and Kairos Power partner to deploy 500 MW of nuclear power for AI

• Databricks and AWS partner to develop custom models with Trainium chips

• Boston Dynamics and Toyota partner on general-purpose humanoid robots

• Accenture launches Federal AI Solution Factory with Google Public Sector

• DARPA plans test of autonomous flight capability on Black Hawk helicopter

• Google announces $15 million in AI training grants for government workers

• NVIDIA stock falls 5% as the US considers expanding AI chip export limits

• New York Times sends Perplexity ‘cease and desist’ notice over content use

• Google introduces NotebookLM Business with Audio Overviews updates

Writer RAG tool: build production-ready RAG apps in minutes

RAG in just a few lines of code? We’ve launched a predefined RAG tool on our developer platform, making it easy to bring your data into a Knowledge Graph and interact with it with AI. With a single API call, writer LLMs will intelligently call the RAG tool to chat with your data.

Integrated into Writer’s full-stack platform, it eliminates the need for complex vendor RAG setups, making it quick to build scalable, highly accurate AI workflows just by passing a graph ID of your data as a parameter to your RAG tool.

Learn more about our production ready RAG tooling here.

📖 What We’re Reading

Rewiring for the era of gen AI

💻️ AI Tools and Platforms

• LatticeFlow → Evaluate AI for EU compliance

• Bytebot → Code-free web automation

• Maxim → Enterprise stack for AI applications

• poolside → Next-gen AI for software development

• Otto → Automate manual research tasks