Trying to keep up with news about AI is a bit like trying to drink from a fire hose. So much happens daily, it will take the benefit of time to fully grasp which news will remain significant in a year.
However, one announcement in October stands out among all of them: President Biden is launching the equivalent of an AI Manhattan Project in the sense that the National Security Memo (NSM) of October 24 will lead to pouring as much as a trillion dollars into ensuring “safe, secure and trustworthy” AI. China has already invested $184 billion in AI, but the US hasn't yet directly invested in the industry. That’s about to change.
US goes all in
The US has been the hands-down leader in AI development, not only because most AI companies are located here, but also because of access to venture capital enabling those companies to buy up much of the AI talent around the world by acquiring other companies. The CHIPS Act, too, has given the country access to 90% of existing AI chips. The subsidized TSMC facility in the Phoenix area is now producing 4% more than its Taiwan-based counterpart, according to a story in Bloomberg Business News.
But when the White House issued the NSM press releases, it essentially confirmed that investing in AI is the top national priority in defense, government and business. This was a quiet announcement with immense implications. The directive established the AI Safety Institute and the National AI Research Resource. The first entity will review AI safety government-wide; the second will provide access to AI research from educational institutions as well as corporate resources and individuals.
AI agents
Announcements about the release of AI agents may sound like yet another new advance in a year of AI learning breakthroughs, but it likely could be one of the most significant developments of the year. AI agents are designed to automate work by executing complex, multi-step tasks, most notably in sales and marketing. They can be trained using natural language. Salesforce announced that it had released AI agents in September, and now OpenAI, Anthropic and Microsoft have all announced plans to create autonomous AI agents.
If you’ve used the programs Canva for graphic design or Notion for business tasks, you may have already been exposed to AI agents that respond to multi-step requests made by typing queries. Imagine having a Siri or Alexa that goes beyond answering questions to become a personal assistant. CNBC.com: “Anthropic said that future consumer applications include booking flights, scheduling appointments, filling out forms, conducting online research and filing expense reports.”
AI slop
This image of a bedraggled little girl and a puppy in a boat tagged to say that it was from Hurricane Helene is one example of AI slop, the kind of web content created to confuse the public with its realism or take advantage of naive internet users to sell products. AI slop tends to be low-quality content that clogs up feeds and is similar to email spam. It’s used for clickbait, poorly written blog posts, images like “Shrimp Jesus,” AI-generated books and AI-generated Microsoft travel articles. In short, it’s dodgy, unreliable, and usually inaccurate rubbish that’s best avoided. Watch out for slop on social media.
AI hype index
MIT’s magazine Technology Review covers the array of bizarro “advances” being hyped as potential and current uses of AI, including everything from creating a “digital psychological twin” to make end-of-life decisions for incapacitated people to creating a digital deepfake twin of yourself to make realistic videos using Synthesia’s technology. DeepMind created a robot that can beat amateurs at table tennis — but not professionals. Much of the ineffectual AI activity will eventually go by the wayside, including Elon Musk’s silly AI creation Grok, due to its lack of rules and guardrails. (It will let you make images of anyone doing anything — not such a good idea.) So much ridiculousness, so little time.
Rebooting Three Mile Island
You know plans for AI expansion are serious when a company buys and relaunches one of the most infamous nuclear power plants, one that had a meltdown in 1978. Okay, this story is from late September, but I felt it was significant enough to include.
Microsoft shows how forward-thinking it is because AI computers are among the biggest energy hogs of our time. That may change in the future, but for now companies invested in AI want to throw everything they can at solving problems and not worry about the one involving their use of resources. The electricity demands are mind-blowing. Expect to see more of this; Arizona has at least eight solar fields in the works that I know of, and probably more to come.
