Not Intelligent

Doghouse – AI

Article from Issue 273/2023

Earlier technologies have persisted despite government regulations – so will AI.

I have been writing articles for Linux magazines for over two decades. Some articles were very technical, some were business-oriented, and some had to do with the philosophy of free and open source versus closed source in producing systems.

However, recent movements by various governments to "limit AI" have caused me to write an opinion article this month describing why this limiting will not work.

I look back in time to other "limiting" exercises, such as embargoes. Do we really think that countries that are under embargo cannot get the hardware and software they need to do whatever they want to do with AI, either by smuggling it out of some country or relying on one of their "partners" to break the embargo?

I remember the limitations on encryption back in the day. The United States has such arcane rules about shipping encryption out of the country that even if you bought a box of encryption software from Canada, never unwrapped it, and tried to send it back, it was cause for the Feds to come after you. We tried to tell the government that all of the best cryptologists were leaving the USA for places where they could sell their skills, but no use – the government would not listen.

Now it is artificial intelligence (AI). The mighty and the wealthy are coming together to tell the software community that the government should license and control the use of AI as if no one had seen even one movie about AI. As if you need a computer the size of a room in order to do AI. As if these titans of industry could somehow put locks on every mind who has ever thought about AI.

First there is the issue that most people cannot even define "AI." They think computers will never become intelligent enough to do their job. That is their first mistake. If you are talking real AI, then the computer (or robot) will not only be able to do your job, but it will do the job faster and better than you, with no time off to sleep or eat – 24 hours a day, 365 days a year. And the computer/robot will not only be able to do your job, but it will build computer/robots like it or better than itself to do your job in parallel.

I have seen AI go from a concept in science fiction to where it is today. For many years, we underestimated the amount of CPU, memory, data, and other factors needed to duplicate human thought and consciousness. Most people still underestimate those factors, possibly by several orders of magnitude, yet through a combination of massive data farms, huge amounts of processing power, and methods of programming, we can automate a lot of the filtering that can bring us closer to what appears to be intelligence.

I am one of those people who think the human mind is a sophisticated electro-chemical machine. Neurons (about 86 billion, +/- 8 billion) supply "storage" and are connected by synaptic connections (7,000 synaptic connections per neuron to other neurons) – sparks of energy that fire off the 5 to 50 "messages" that each neuron sends per second to other neurons in ways that we still are exploring. In an adult, these synaptic connections measure in the trillions.

There is a certain amount of "programming" built into us through eons of evolution, but a baby is still in some ways a blank slate and is influenced and taught how to think – to grow those synaptic connections.

Yes, we are amazed at what ChatGPT can do, but remember that many of these examples have been shown to be far from "intelligent." They come from datasets which may not be complete, and which have been shown to be inaccurate in ChatGPT's answers. These examples often need context which is outside the dataset being used by ChatGPT, even as huge as it is.

All of that having been said, someday AI will be "intelligent," and no laws will stop its exploitation or misuse.

The governments and companies that try to limit its development will only force other governments and companies who want to "get ahead on AI" to go underground. We will have an AI race just as we had a space race and a nuclear arms race years ago. Except you do not need to shoot off rockets or ship uranium to develop AI.

We are no longer in the realm of computer science where you need billions of dollars and government support to build a system. Almost any individual could fund the development of an AI system, just as they wrote their own encryption code years ago. Each developer will justify their development outside of the law.

The only way to win the war of AI is to engage in it openly. Then you may be ready when AI comes to your doorstep.

The Author

Jon "maddog" Hall is an author, educator, computer scientist, and free software pioneer who has been a passionate advocate for Linux since 1994 when he first met Linus Torvalds and facilitated the port of Linux to a 64-bit system. He serves as president of Linux International®.

Buy this article as PDF

Express-Checkout as PDF
Price $2.95
(incl. VAT)

Buy Linux Magazine

Get it on Google Play

US / Canada

Get it on Google Play

UK / Australia

Related content

comments powered by Disqus
Subscribe to our Linux Newsletters
Find Linux and Open Source Jobs
Subscribe to our ADMIN Newsletters

Support Our Work

Linux Magazine content is made possible with support from readers like you. Please consider contributing when you’ve found an article to be beneficial.

Learn More