Bet You Didn't See This Coming
You've probably heard about the ways in which AI or artificial intelligence is going to change our world. Those that stand to gain the most financially from its development extoll its virtues and claim that a veritable utopia awaits. Those who are more circumspect, including Elon Musk among others, worry about all the ways it could go wrong. (See my Books page for a brief overview of "Empire AI.)
I am part way through "Scary Smart, The Future or Artificial Intelligence," by Mo Gawdat. Though I'm not finished I am already stunned by what I have learned. The man is no slouch, nor is he a new-comer to the world of AI. (His site is linked at the end of this post.)
Part of his bio from his web site:
"With a career spanning 27 years at IBM, NCR, Microsoft, and Google, Mo helped launch close to half of Google’s global operations and later led business strategy at Google [X], home of projects like self-driving cars, Project Loon, and Makani. Driven by a passion for solving humanity’s biggest challenges, Mo has cofounded more than 20 companies and mentored countless startups. "
He speaks with authority on the topic.
One of the things he stresses is the fact that computing power is increasing exponentially and at an incredible rate. (You've heard that before, but hold onto your hat.) Today's AI runs on chips known as GPUs, initially invented for applications that require high graphics performance, such as video games or generating video sequences. Nvidia is one of the biggest makers. AI companies use these because they are among the fastest and most powerful available. A data center might have as many as 10,000 or more of these running simultaneously.
That's a lot of computing power but the only way that companies (OpenAI, Anthropic, etc.) can expand the capabilities of their AI offerings is by building more data centers and adding more and more processors; data centers, that require more electricity and more water to cool those hot CPUs. There is already opposition to data center construction in cities and towns across the USA and the world.
But there is a new sheriff in town. Its name is Quantum Computing. Traditional computers can hold only for states of two options: off or on, 0 or 1, the binary limitation. Bit pairs can either be 00, 01, 10, or 11. Quantum computers can hold all four of these possible combinations at the SAME TIME.
Google's new Quantum Computer Sycamore can hold more than 10 quadrillion combinations. (10,000,000,000,000,000.) And, no, this is not science fiction, it is the emergent computing model. Imagine how much faster it can perform.
When the AI industry moves to Quantum Computing we won't know what hit us, things will happen so fast. The holy grail known as Artificial General Intelligence or AGI will arrive in an instant. AI will be self learning, capable of making decisions on its own, able to make infinite copies of itself, and it will be a billion times smarter than humans. No one really knows what "that" is going to be like, but it is coming, and it can't be stopped. There is just too much money behind it, and the geeks of the tech industry want this.
