Although I have made good use of Artificial Intelligence on a few projects, I don’t consider myself an expert on AI in its present forms. However, I’ve been familiar with the basic concepts of AI since the late 1960s when I was a graduate student at Carnegie Mellon University.
I wrote my PhD thesis on one of the first Artificial Intelligence programming languages, “Information Processing Language – 5” and two of my PhD thesis committee advisers were pioneers in artificial intelligence.
Dr. Allen Newell created IPL-5 (information Processing Language-5) and wrote one of the very early and most effective chess-playing programs. This was used as an example of a computer trying to think like a human.
Another advisor, Dr. Herb Simon, won a Nobel Prize in 1978 for his work on the decision-making process within economic organizations. He and Newell also received the distinguished computer science Turing Award in 1975.
Therefore, I’m at least very familiar with the concept of using computers and software to mimic the capabilities of the human mind.
My Investments in AI-Focused Companies
Several of my venture capital investments have intersected with this concept and Artificial Intelligence.
In the early 1980’s, Speech Systems, Inc. was developing a natural language speech-to-text translation algorithm. I also invested in Adaptive Solutions, Inc., a Portland, Oregon company making a neural network on a chip, which is one of the models for the way human brains function.
One of my current projects is developing an extremely high-powered, adaptable computer engine optimized specifically for Artificial Intelligence programs.
So, maybe I know enough about artificial intelligence to be “dangerous”?
The Growth of Artificial Intelligence
The basic concepts of how to simulate human intelligence may not be very different today. AI programs have historically used concepts like pattern-matching, neural networks, and exhaustive evaluation of alternative solutions to problems. But aside from large language models, I don’t feel very knowledgeable about the present algorithms.
What I do know, is that the largest driving force behind AI is probably the ever-advancing power of computer technology. Since 1968, the increase in computer power is something like 7200 trillion times. Which is just amazing!
This ever-increasing computer power enables simple algorithms to be repeated countless times, allowing them to mimic human thinking.
Increasing Computer Power Fuels the Capabilities of AI
In just the twelve years following my leaving Carnegie-Mellon University and Allen Newell’s chess-playing game in 1980, chess-playing computers evolved from machines large enough to fill a room to devices the size of a pocket calculator. This represents an approximately 4000-fold increase in computer power!
A device the size of today’s cell phone can play chess almost as well as a Master. Moreover, today’s chess-playing programs almost always leave the Masters in the dust.
Similarly, I watched computer power evolve from the point of voice recognition software running extremely sluggishly on the largest computers of the early 1980’s, to now being implemented on an iPhone with amazing accuracy.
Have there been improvements in the algorithms? Probably. But I think most of the improvement is due to increased computer power and the ability to evaluate millions of possibilities in microseconds.
How Will Artificial Intelligence Be Used in the Future?
Can AI be a good thing? Absolutely! Advances in voice recognition, robotics, and sentence completion are astounding.
I am certainly not an AI power user, but I have used it dozens of times to identify solutions to problems with good results for tasks such as creative writing, writing complicated passages of papers, and suggesting ways to solve a medical problem. I’ve even used it when making suggestions for a particular type of meditation. These are minor uses. But they are still well beyond the capability of today’s search engines, which suggests that Artificial Intelligence will very likely encroach on many functions that humans perform today.
I have to believe, for example, that much of the writing for marketing and advertising departments today can be effectively done by artificial intelligence.
Also, I’m certain that AI-driven robots are becoming more and more effective at what they do.
Here’s an example: When I was at TRW in 1981, I watched the demonstration of an AI-driven robot, whose task was to identify three letters, T, R and W, which were scrambled on the table, and arrange them into “TRW.”
The demonstration turned out to be pretty hilarious, because Simon Ramo, the “R” in TRW, was in the small audience, and the robot kept dropping the “R,” much to Ramo’s disgust. Somehow, I don’t think today’s robot would drop the “R.”
Here are some of the first jobs Artificial Intelligence is expected to take over, though the trend will move toward more sophisticated tasks.
- Data Entry and Administrative Tasks. One of the first job categories in AI’s crosshairs is data entry and administrative tasks.
- Customer Service.
- Manufacturing And Assembly Line Jobs.
- Retail Checkouts.
- Basic Analytical Roles.
- Entry-Level Graphic Design.
- Translation.
- Corporate Photography.
Source: Bernard Marr on Forbes.com, June 17, 2024. Accessed July 31, 2024.
What Will This Mean for the Future?
In past technology revolutions, technology displaced relatively mundane jobs but provided opportunities for more advanced or more sophisticated ones. Will that be the case with Artificial Intelligence? Possibly, but it’s hard to know since AI gradually taking over more sophisticated tasks is the trend.
Can the future of AI be bad? Maybe. Artificial Intelligence can have a bad or even devastating influence on society in many ways.
For example, the extent to which people rely on AI as a sort of search engine. The results can be intentionally biased in ways that are very difficult to detect. We have already seen some examples of this in the form of clearly biased political responses. Currently, the only real protection here is caveat emptor – buyer beware.
One of my greatest concerns is the use of Artificial Intelligence as a military tool. To me, the standout example of the ultimate danger would be swarms of gazillions of bug-sized, lethal, flying, AI-controlled robots. How do you defend against that? The only real defense seems to be an even more frightening counter-offense.
Can, or Should, AI Be Controlled or Regulated?
There is probably an endless list of reasons that AI should be controlled or regulated. For example, it shouldn’t be used to defraud people of assets. Yet some of these regulations may simply fall under our existing legal system.
But I think it will be very difficult to regulate Artificial Intelligence. One reason is proliferation—very powerful systems can be built by a very small group of people. So, how do you know whom to regulate? And, how do you enforce regulations?
Regarding international threats, how can we possibly know what our enemies are doing in their AI labs, much less regulate them?
It’s not a very happy answer, but I have a feeling the only real answer is to out-AI the other guy—to develop AI systems that are so powerful that they can detect and outsmart threatening AI approaches being planned, developed, or used by our enemies, both within our country and without.
In conclusion, this suggests that we should put a full-court press on being the world leader in AI technology!
These are just some of my thoughts. What are your thoughts on Artificial Intelligence and its uses?
Fred, your article offers a fascinating look into the evolution of AI and its increasing role in various industries. As someone who has been using AI for the last couple of years, I agree that advancements in computing power have played a critical role in AI’s growth. While its potential for innovation is vast, the concerns around regulation and misuse are equally important. I believe your idea of a balanced approach, prioritizing both development and safeguards, will be essential for responsible AI evolution.