Too much intelligence? When do we say enough A.I.?

I had an opportunity this past Sunday to make a presentation on artificial intelligence and data centers at First United Methodist Church in Madison, Wisconsin, and I think that the importance of this issue makes it worthwhile to share my talk more broadly. As we perhaps all know by now, data centers are huge consumers of electrical energy and water. Proposals to locate so-called “hyperscale” data centers in communities typically and increasingly are met with local opposition. People are beginning to understand that the promised construction jobs created for local people in building these monstrous complexes are few and quickly fleeting, long-term employment for the locals during the operation phase is almost nonexistent, but the draw on the electrical grid equivalent to tens of thousands (and possibly even hundreds of thousands) of households will remain forever.

My talk (which begins at about 2:25 in the video, following an introduction by the moderator) was the second in a two-part series relating to hyperscale data centers and artificial intelligence. The first speaker, on May 10, had been UW Professor Anna Haensch, a mathematician and data scientist, who spoke especially about hyperscale data centers generally and as proposed (and largely opposed) in Wisconsin. There are three hyperscale data centers going up in Wisconsin now, several more proposed, and some more that have been defeated by citizen action. Haensch’s talk really clarifies things a lot and was really excellent. I am including a link to her talk here.

A data center is just a collection of dozens, hundreds, or thousands of computers storing data and providing computation for businesses or other organizations offsite. Very importantly, Professor Haensch differentiated between three types of data centers, only the latter of which, she noted, warrant public attention:

1) Colocation facilities, which typically contain several dozen computers serving local businesses are “not a big deal,” she related. “These things do not impact our lives whatsoever.”

2) Enterprise data centers, which form a second tier of data centers, are much larger, perhaps covering a couple acres, and “do not really strain the (electricity) grid…You would never think twice if you saw (one)…and it’s not bothering anyone.”

3) Hyperscale data centers constitute the type that is “taking up by far the most oxygen, literally and figuratively.” This, she explained “is an entirely different beast,” each taking up hundreds of acres, and requiring massive amounts of electricity. This is the only type of data center that people need to think about, Haensch said.

My task was to speak about artificial intelligence (A.I.) generally and in relation to data centers and the environmental and community effects. (Originally my talk had been scheduled for May 17 and had actually begun, but all the audience’s mobile received tornado warnings about 5 minutes into my presentation, so the audience went to the church basement and we rescheduled.)

At the end of my presentation, I drew the following 10 conclusions:

1. First of all, the argument will be made that artificial intelligence is the way of the future and so we need to have all these data centers. That is not, in my opinion, a correct characterization of the situation. Notice that these are called DATA centers, not INTELLIGENCE centers. From environmental and privacy perspectives, the main problem is data. Huge amounts of data have been collected on every one of us and that collection continues even as it grows exponentially. Nobody wants to delete anything; they only want to pile up more.

2. Human beings, including all of us, have lost control over our own data. We have lost control over privacy. We have no privacy anymore. Europe has tried to control the collection of private data, while the U.S. has not. To be honest, I think the result so far is probably about the same in either case. If, as a matter of public policy, we could restore personal privacy, if we could take back our data, then there could a lot less of it to store, process, and exploit.

3. A.I. and especially social media should be regulated – and the companies that operate these services should be compelled to bear the full economic, environmental and social costs of their activities. This will be difficult – and maybe impossible by this point – for any number of reasons. The ship has already sailed on social media and the A.I. horse is already long out of the barn. I do not know the answers on how to do this. That is a job for public policy experts who are much better informed and smarter than me. Unfortunately, we have a political environment now that fights against any sort of public policy. But there should be regulation to protect privacy, to protect children and teenagers, and it should hold the Mark Zuckerbergs, the Jeff Bezoses, the Bill Gates and other billionaire oligarchs fully and legally liable for whatever damage they do – no exceptions.

4. Data and data centers constitute a huge environmental threat. At minimum, they should be forced to pay much higher prices for electricity than they currently do. In my opinion, they are robbing the electric grid. Better still, they would be prohibited altogether from taking electricity off the grid for crypto mining and for A.I. that is used for purposes not demonstrably in the public interest – and that, I think, is most of those purposes.

5. Perhaps, the technology companies might be allowed to produce their own power – as many of them are now planning to do. For example, Bill Gates has proposed to restart the Three Mile Island nuclear power plant that had partially melted down in 1979. He’d like to have taxpayer-funded loans to pay for it and tax credits to make it profitable. I wonder when we’ll see somebody propose to restart Chernobyl and who that will be, Elon Musk maybe? If I were King of the World, I would prohibit technology companies from using fossil fuels or nuclear power for their purposes.

6. The technology companies should be required to cover the full costs of their activities – including the externalities, which damages that they do, environmental and otherwise – and reflect those costs in either a) the prices they charge to their end customers, or b) in lower profits for their shareholders.

7. The penalties for using crypto, A.I., and/or private data for illegal purposes should be strong. We have fines that double or triple for speeding in school zones or construction areas. Punishment for armed robbery is more severe than if no deadly weapon is involved. Punishments for crimes committed using crypto could be, I dunno, 10x more severe than if crypto is not used.

8. Everybody needs to recognize that what we are witnessing here is unbridled greed at work. Massive, hyperscale data centers will continue to be built – contrary to the public interest – because political actors will be bought off. There is too much money involved here for that not to happen.

9. The United Methodist Church, ecumenical associations, and other religious denominations and organizations need to take a stand and to act. Pope Leo has shown leadership in this area. There should be room for ecumenical action.

10. Finally, I would pose the question where and when should the general public oppose massive data centers? In my opinion, the answer to that is everywhere and always. Whether or not we oppose any given data center, there will be enough capacity for the uses of A.I. that benefit society and planet and too much capacity for the things that do not.