AI Alignment, Politics and Big Tech

The illustration above seems to be the obvious solution to the AI alignment issue. But the big question here is, who is “us?”

I am interrupting my current post series on tech studies due to the latest news from the current administration regarding AI Policy. All of the ideas contained in this post are mine and mine alone.

During the Coronavirus pandemic worldwide, education faced myriad challenges regarding teaching and learning. Of particular concern was grading. With students unable to sit for their A-Level exams (similar to our, SATs) a solution was offered by a school district in Great Britain. They used a computer algorithm to predict what the students would have received for a grade based on data from previous years. The results were disasterous. The model favored students from private schools and wealthier districts. Higher achieving students from free and state schools scored disproportionately lower, leaving them without college placements and job prospects. Education leaders quickly responded with an apology and rectified the situation. Lesson learned. Maybe.

We are faced by many pressing issues with the growing use of AI and its capabilities for exponential computing power. At the end of the infrastructure chain, AI is powered by fossil fuels and it’s a very hungry animal. The minerals required for digital components are scarce and found in countries that are engaging in brutal, extractive economies, outright slavery and warfare for domination of these resources. At the end of their use cycle, poisonous materials often end up in landfills in these very same countries. With all these concerns aside, the problem of “alignment” is possibly the most critical one at this very moment. The example of Great Britain above is likely to be played out over and over and in a legion of ways.

Then we have the various groups with gradations of AI acceptance, disbelief, blissful hope, rejection and depression. Which one are you in? The blissful hope segment believes that AI will solve the climate crisis and usher in the long awaited utopia. Rejectionist will go “scorched Luddite.” The acceptors are the type of people who say, “well, you gotta die of something.” The depressionists see a bleak and total dystopia just beyond the horizon.

Al Gore was one of a handful of US leaders that actually “studied” science. He once asked a group of computer specialists what it would take to build a computer that could account for all carbon emissions on the planet. The answer was that such a computational infrastructure would, itself, be the largest carbon producer that it’s model would measure!

But what do our politicians see? More importantly, how much do they understand this technology or for that matter any technology? I can confidently conclude that the current president of the United States simply does not have the intelligence needed to create policy on this and his staff are likely just as uneducated as he is or simply blindly compliant.

Those of you reading this that support our current administration need not worry. Here is the language from the 28 page action plan itself. In a rollback of Biden’s AI rules the manifesto states it will “eliminate references to misinformation, Diversity, Equity, and Inclusion, and climate change.” That’s right. Bingo. “Just say no” to climate change. He also said this “The American people do not want woke Marxist lunacy in the AI models,” as he signed an executive order entitled “Preventing Woke AI in the Federal Government.” Diversity, equity, inclusion? Wiped.

The graphic above makes it easy to “follow the money,” and see who the players and influencers are and what they stand for.

The tech oligarchy have patiently waited in line for their chance to kiss the ring at million dollar a plate dinners. With Musk on the outs, Sam Altman rushed in to fill the void. A longtime democrat, he once called the president a “threat to America.” Now he sees him as the de-regulation angel he’s been praying paying for and has disclosed “no party affiliation, might vote republican” at the moment. The president is promising a laissez-faire approach to AI big tech and intends on punishing states that institute serious regulations on AI. And Altman is no genius either. Nor Musk. None of them are. They are clever, rich, powerful men and they want to stay that way. The emperor may have no clothes, but he does have the most powerful men in tech on his side, helping him build a version of his truth, his vision, his reality. The goal of AI alignment is to account for and mitigate bias. When thinking of what should we align AI to, the worst possible choices are these.

I suggest we establish yet another new political party. I call it the “lickspittle” party, there are enough toadies and sycophants in this administration to fill a Qatari Jet.