Practical AI

From Fig Newtons to Big Newtons

• By Howard M. Cohen
• 04/15/2025

Remember Fig Newtons? No, they weren’t cookies… they were fruited cake!!!

Then, on August 3, 1993, came the Newton, introduced by Apple Computer as the world’s first personal digital assistant (PDA). Conceptually, of course, this was inaccurate. A full quarter-century earlier, in 1968, visionary futurist and computer scientist Alan Kay first envisioned the KiddiComp concept which led to his proposal in 1972 of the Dynabook as "a personal computer for children of all ages." Note the date. Kay’s proposal preceded Apple, IBM, and anyone else who would like to take credit for the phrase "personal computer."

Kay was aware of Vinton Cerf’s team busily developing ARPANET, which would become DARPANET when the Department of Defense took over the project, and ultimately the internet we know today. He envisioned the Dynabook being able to connect to all the knowledge in the world through this new global network concept. It would enable incredible computing and composing power in the palm of your hand.

Many remember how, 25 years later, then-Apple CEO John Sculley demonstrated the handwriting capabilities of the Newton at a convention and then sat down to listen to the next speaker. When he took out a paper pad and pen to take notes, the crowd crucified him. Why didn’t he use the Newton? I’d be willing to bet he still ponders that question today.

The Little Newton

The Dynabook, the Apple Newton PDA, and all the PDAs that have followed it hold the promise of putting a world of information and productivity into the palm of your hand. Indeed, many people today do the overwhelming majority of their information gathering, communicating, and computing on a hand-held device.

We can look at all these descendants of the ill-fated Newton to be Little Newtons That Could.

Welcome Big Newtons

The core concept of the personal digital assistant has now morphed into much larger form. We can readily create them using a natural language interface to tell them exactly what we want them to accomplish, under what conditions, with great specificity. We call them "agents."

Agents may be accessible from our handhelds, but we default to thinking of them as being a product of our computers and the networks they run on. When we craft them correctly, we can ask any of our agent "PDAs" to do what it was built to do any time we need that thing done. Instead of having one all-purpose PDA like the Newton, we have many, many PDAs at our disposal. Much as we have many fruited cakes in our package of Fig Newtons. And we can access them from just about anywhere at any time.

Gain Solace from my Metaphor

Notice that many people are employing metaphors to explain how AI does what it does and why. This is my friendly Fig Newton metaphor. Agents are the PDA of today, unbound from their hand-held limitations and fully integrated with anything we care to integrate them with.

If you’re among those who are fearful about the "rise of the machines," I hope you’ll find solace in the idea that these things work for us, doing what we tell them to do.

Artificial Intelligence – From Generative to General

However, this may not last forever. Many computer scientists today talk about the day we achieve Artificial General Intelligence (AGI) not to be confused with Generative AI, which is software that produces things based on input plus what it has "learned" during training.

AGI happens when computers can match or surpass human cognitive and intellectual capabilities. The brightest minds, including Bill Gates, Stephen Hawking, and others have expressed their concerns about what happens when AGI is achieved.

American computer scientist, author, entrepreneur, futurist, and inventor Ray Kurzweil, goes even further to describe "the singularity," the point at which human and machine intelligence merge. His theories go well beyond what we're seeing at Neuralink, the company founded by Elon Musk to develop brain-computer interfaces (BCIs) to help people with neurological conditions, such as paralysis, and potentially enhance cognitive abilities.

This leaves us to ponder the many ways in which integration between people and their machines may be achieved.

Everything is Derivative

What we have today is generative AI, machine-learning, and something approaching machine reasoning. All of it is based on a computer’s ability to scan vast amounts of information and make the best estimation of what comes next when creating anything from text to music to images—or computer code. We can even put our thumbs on the scale by adjusting the "weight" various concepts are given. They’re not thinking; they’re forecasting based on data.

One of the most derisive things you can say to any creator is that everything is derivative. That is, everything they claim to create is simply a re-hashing or re-arrangement of information they’ve encountered before. "There’s nothing new under the sun!" is like nails on a blackboard.

This may be true, but for today it seems to be what we want and need. For decades now, our technological progress has far outpaced our political logic. Buckminster Fuller, the American architect, systems theorist, writer, designer, inventor, philosopher, and futurist, famous for his geodesic domes, may have told us that anything nature lets us get away with must be considered "natural," but we are approaching a moment when we can invent something we may not be able to control.

You’ll read many arguments for and against moving forward with the further development of AI. Some will say we can’t let China beat us to it or suffer their supremacy for decades to come. Others say we put our civilization at risk.

For now, I don’t know who’s right, but we do currently have something we can put to incredibly good use in many, many ways. What would you like to teach your agent PDAs to do for you?