My first major job after college was working for an organization called the Computer & Communications Industry Association in Washington. It was basically a policy and advocacy group for computer and regional telephone companies that lobbied for policies that would promote the confluence of computers with phones.

Back in the late 80s, the assumption was that when the inevitable wide-scale connection of computers with telecommunications networks happened, the phone companies would be the big winners because they controlled the wires.

But there were other players like CompuServe, Prodigy and AOL that got to that space first. And around 1990, I started to write about this thing called ARPAnet, a network of government agencies and universities, that was starting to gain traction. It soon would become known as the internet.

I was writing newsletters and conference briefing books at the time on way out subjects, like how television one day would migrate to computers. No one at the time anticipated that wireless technology would take over and that computers and phones would literally converge into a handheld device.

Given how the technology could have played out, we were extremely fortunate that the U.S. government was investing in this technology backbone that became the internet and was basically willing to give it away for free. The more commercial information services that came first were far more limited and locked users into a particular piece of software and account.

This is very similar to where AI is right now, in numerous apps with their own strengths and drawbacks. But unlike what happened in the early days of the internet, there is no government built and supported infrastructure in place that could potentially open up AI for all.

Yes, we all saw the internet create vast wealth for a handful of companies, but it didn’t start that way. Many of the big early players — Netscape, AOL, Yahoo — are now gone. Google at first seemed like just another best search engine, taking the place of AltaVista, whatever that became. Companies rose and fell with innovation and consumer preference.

That can’t happen with AI. This is an arms race right from the start. And I find myself in two minds about it. First, having seen the birth and growth of so much technology in my lifetime, I remain generally optimistic about its value and potential. Technology can and should make people more productive and efficient — freeing them to toil less.

But what concern me is that we’re in a race where no one knows the outcome or even the goals of the technology — only the desire for world-shifting wealth. So theft is taking place at an unprecedented scale. People are getting addicted to chatbots for no reason other than a programming desire to hook people with sycophantic praise and guacamole recipes.

In short, I think we had a much clearer vision of what the internet would become in 1990 even before we knew its name. Today, we’re just building black boxes in a massive hurry — not caring how much intellectual property we plunder, how much energy we consume, how school essays are ruined and realistic self images warped. And we’re still in the early stages of the deep fake revolution and all that will do to the culture.

When I see someone like Peter Thiel bemoan that we don’t have flying cars and colonies on Mars and wonder why human technology slowed down, I have to fight the urge to just get angry and for a moment consider: what if he’s right about that. If what Thiel believes to be true, and it is the current state of humanity, is that a bad thing?

The answer to me is clearly no. The mid 20th century brought us the incredible trauma of a global depression, another world war, mass genocides and the invention of nuclear weapons. It is a sign of great intelligence for humanity, in a moment like this, to consider the need to slow down, to work on humanity’s moral improvement, otherwise we could easily race to annihilate each other.

Whether that slow down ever happened over the last 75 years is debatable — it’s certainly not true in the tech world and Thiel concedes this. But I believe it’s become apparent that as AI is concerned, it’s becoming a necessity to hit the pause button.

What does an AI pause mean?

First of all, we need to let the bubbles pop. The valuations of AI and AI related companies right now are insane. They will never come close to paying off — people are going to lose a lot of money in this gold rush and we need to let it happen. There will be consolidation as a result, that’s not necessarily bad.

Second, we need regulation. If it can’t come from the U.S., then the EU will have to step up and demand it in the same way they have regulated privacy rights. Safety standards have to be built in. Some clear understanding of what Artificial General Intelligence is has to be established before we race toward it. And we have to demand a kill switch for everything in case AI starts making decisions counter to human thriving.

And third, we need a mechanism in place to redistribute income before it becomes super concentrated. Because, if the promises of AI come even close to fruition, a tiny number of players are going to hold wealth unimaginable by today’s standards.

It’s too late now to do AI right — to bring it back inside true nonprofit organizations (what openAI was intended to be) or universities. The best we can hope for is an AI with just a sliver of control.

Otherwise, we’re doomed to roll the dice on a commercialized Manhattan Project to create god in a box. And we wonder why intergalactic aliens keep a safe distance from earth.