Everyone is talking about Google $GOOG. They’re making big moves and getting investor love.
I’m suspicious of companies trying too hard. At its November 26 opening bid of around $365 per share, the market cap of the stock is now 32 times annual earnings. This brings it just about even with Microsoft $MSFT, which was the general idea.
In the AI Bubble, market cap is your superpower. The more your company is worth, the more it can spend on new data centers, and the more it can borrow. Google’s $25 billion in debt looked substantial early this year, when the company was worth $1.8 trillion. At $3.6 trillion, it’s nothing. Google’s bonds are now priced to yield about 5%.
But this story is about Microsoft
I was writing about the difference between Big AI programs like ChatGPT and Little AI programs over two years ago. Big AI wants to do everything, Little AI just one thing. An example might be a customer service AI, or a personal agent loaded with just the data and sites you use or have used.
For Little AI, enterprise software companies like Salesforce.com $CRM and ServiceNow $NOW were the canaries in the coal mine. Those canaries have grown ill in the last year. They’re both down over 20%, despite brokers “pounding the table” for them.
The difference between Big AI and Little AI is that Big AI lies while you can train Little AI not to. I think Little AI is more sustainable. The stock market prefers the “Great Game.”
Fine. So why is Microsoft suddenly interested in Little AI? Especially since that’s not what investors are looking for right now?
Microsoft’s New Direction
Mustafa Suleyman, now CEO of Microsoft AI, laid it out in a blog post. Since such blog posts are marketing, he has a buzzword. It’s Humanist Superintelligence, or HIS. “Not an unbounded and unlimited entity with high degrees of autonomy,” he writes, “but AI that is carefully calibrated, contextualized, within limits.”
Sounds like Little AI to me.
“We have to ask ourselves, how are we going to contain (secure and control), let alone align (make it “care” enough about humans not to harm us) a system that is – by design – intended to keep getting smarter than us? We simply don’t know what might emerge from autonomous, constantly evolving and improving systems that know every aspect of our science and society.”
Microsoft believes this approach can yield AI Companions in education, Medical Superintelligence for diagnosis and research, and clean energy solutions to power data centers. Note that these aren’t little jobs, but they are narrowly defined, the software designed to interact safely with their users.
The political aspect to all this is that Microsoft welcoming regulation, something that’s anathema to its rivals and the current Administration. “Accountability and oversight are to be welcomed when the stakes are this high,” Suleyman writes.
Everyone in AI knows there’s a cliff, a market crash, ahead of us. Everyone knows that LLMs don’t deliver Artificial General Intelligence and likely never will. Microsoft, like IBM $IBM a generation ago, wants to be around on the other side of that cliff.
From what I have learned about tech history, I wonder if they will be.
