The story of AI in a few decades might be the story of “good guys always finish last” as shown by the rather unedifying spectacle of the corporate coup at OpenAI. Idealistic and altruistic chit chat are all very well but are trumped by deeds and money. Helen Toner’s rumoured ethical concerns have achieved nothing other than to make her position untenable, not to mention economically unviable.
The problem is that a lot of people talk, but there are fewer doers, and the doers tend to be driven by motivations like money and power. Indeed, the makeup of the new OpenAI board tells you exactly what you need to know about the newly prioritised values of the board! On it sits an ex-Chief Economist of the World Bank Larry Summers and the CEO of Salesforce, Bret Taylor. Harvard, Stanford and Microsoft have all stamped their authority on OpenAI in just a couple of days, they should really change their name to CapitalAI. The direction under Sam Altman can be summarised by what was on offer at the recent keynote, I don’t need to elaborate any further.
AI is too powerful and too important simply to leave to corporate profiteers.
The dishonesty of it also rankles with me. When those like Altman are out there in public defending the safety and non-profit research principals of OpenAI, to pretend that they are still doing AI for the good of humanity when actually like all companies their goal is survival. OpenAI’s survival depends on billions of dollars of investment and compute power from American blue-chip corporations. Altman has an aptitude for putting up a smoke screen in public with lots of lower case letters, smilie emoticons and red hearts in statements, but deep down beats the heart of a passionate business person.
Indeed we are all (nearly all) incentivised by profit to gamble and reap the short term rewards of our risks. It is the stuff of Ayn Rand. It is the local neighbourhood butcher motivated by survival and profit (self-interest) that is the most altruistic of all of us because he feeds the neighbourhood, but it doesn’t stop the fact that he slaughters animals for profit.
AI has value as a creative and scientific tool, but it is in competition with other forms of intelligence such as our own. With globalisation, a global marketplace and the world wide web, do average humans need any more competition? There’s a point where a lot of us just won’t be able to compete and we are left behind.
What’s also interesting about the OpenAI drama is the difference in how men and women regard risk. The women on the board who were for a slower and more thoughtful approach to AI came into direct conflict with male leaders such as Sam Altman, whose calculous of risk is completely different. He isn’t there for a safe approach and one of the reasons the world is both simultaneously such an achievement and such a mess is because people, mainly men, are compulsive gamblers.
One could charitably describe Altman as another Zuckerberg or Jobs, a tech pioneer not afraid to gamble and throw people under a bus in order to get what they want. Ironically it was Larry Summers as Harvard President who allowed Zuckerberg to get away with stealing the idea for Facebook from his fellow students.
These gamblers are technological optimists and evangelists, unlike OpenAI chief scientist Ilya Sutskever who has more of an engineer’s pessimistic mindset, which is good for identifying problems in order to solve each one in a systematic way. Altman is a much more idealistic person and probably actually believes his work will be to the benefit of all humanity!
His goal is to create Artificial general intelligence (AGI) not just because it will make him an awful lot of money. He has nearly all of his employees onboard with this mission as it will also make everyone on the team, especially the first 1000 employees, extremely wealthy too.
During the chaos of Sam Altman’s sacking, his hiring as a Microsoft employee and his short-lived replacement as CEO by Emmett Shear, was a fascinating holding pattern. Microsoft acted fast to avoid competitors taking all the key people. Shear came into steady the ship on behalf of Helen Toner, and find out why Altman had been sacked, but the board left him as oblivious to the reasons as the rest of us.
For me this is unforgivable.
Those responsible for firing Altman have to come clean and tell us exactly their safety-related reasons. They are still being obtuse at the time of writing. What is the point of a silent whistle-blower?
If you have safety concerns or commercialisation concerns, and make such a big move, you have to able to immediately explain it to your customers and investors. The most astonishing u-turn was by Ilya Sutskever who if there is ever a Hollywood movie about the origin of AGI will in all likelihood be played by a bald Cillian Murphy. For all his fearful ramblings on record about the dangers of AI and AGI, he pivoted between two polar opposite philosophical standpoints in the space of 3 days without even bothering to explain to us why.
All in all, the event will be remembered as the moment Microsoft pretty much fully acquired the IP and all the key people behind OpenAI and thus Artificial general intelligence for the price of a minority shareholding in a start-up and lots of compute. Well done to them.
You could speculate this was all too good to be true in fact and it could have been a bit of an inside job or Microsoft-orchestrated conspiracy to own OpenAI outright without that pesky non-profit board getting in the way. Nobody needed the governance change more than Microsoft, to protect their billions.
All in all it sets a very bad example – that AI safety concerns and a more cautious approach will always end up losing to the venture capitalists and corporations who stump up the money in anticipation of massive returns.
This raises an important question, if AI was PROVEN to be an existential risk for our society or even humanity as a whole, or was causing society to descend into civil war, disorder and mass unemployment, which people, which entities, which companies would then have the ability to hit the big red STOP button?
The answer of course, is nobody.
OpenAI’s chief scientist is on record many many times warning about his own technology, as many other AI scientists and researchers are. Anyone with half a brain can see where it’s going. There are enormous upsides to AI in the form of scientific research, technological development, chipmaking and medical cures, but enormous potentially world-ending downsides, and huge implications for future employment, jobs and mental health, as well as it being an accelerant of a growing gap between the have and have-nots.
Emmett Shear is on record saying AI can supercharge the self-replication of itself and of the machines that power it. Microsoft do not want people like this in charge of development, slowing it down, being bad for business. Only optimism and a can-do approach makes money.
These high-minded ethical positions have so far served only to jeopardise positions and strengthen the hand of investors who want to charge full steam ahead without thinking about the unintended consequences.
If a senior figure of a leading AI company in the US says “stop” or even just “let’s slow down and think more” it is clear that the majority of their staff would not be on side either. Many would just get a job elsewhere like at Microsoft or Nvidia.
It is clear if the US government would say STOP, other countries like China would simply power ahead instead.
Individuals are powerless, companies are powerless, governments are powerless, countries are powerless.
The AI arms race is like nuclear armament. Everyone knows it’s a dangerous idea but that hasn’t stopped 80 years of nuclear weapons technological development. Indeed, if your country sat out the Cold War and doesn’t have a nuclear deterrent, you’re still in the shit with the rest of us when the missiles start flying aren’t you? So what difference does your pacifist stance make then?
Since regulation and laws are inevitable, the AI industry has learnt from the social media industry that it is better to work with governments to regulate themselves and have a hand in determining what the rules and guide ropes should be, rather than have rules arbitrarily imposed on them one day, which might be really badly thought out and bad for profits.
There might be a thousand big red buttons we can press when it comes to technology but none of them work. They only push the genie into a different building. Ironically the only big red button that does work is the one in the suitcase linked to a nuclear arsenal. You wouldn’t want to press that just to silence Clippy.
Suppose after 90% unemployment, a Western democracy gets cold feet about how far AGI technology has gone, long after it has bolted free from the gates, and suddenly imposes the most strident regulation imaginable to slow down or even reverse further development and commercialisation. This democracy then has to say to communist Russia and China, please close down your entire AI industry as well.
They will take this as an act of war.
This is where the planet and indeed humanity itself has a major problem, because a safe AI is a dead AI. A non-profit open source AI is an obsolete AI. An ethical AI company is a bankrupt AI company. A country and government paralysed by regulation of AI is a country and government left behind by AI.
After all this comes down to the very nature of what intelligence is all about.
Intelligence is about the survival of the fittest and Darwinian Evolution.
Any attempts to curtail intelligence and the evolution of it is doomed to fail, any attempt to bend it to more benevolent ends or to stop it competing with our own intelligence is doomed to failure. The moment you do this, you’re overtaken by someone who is profit and goal orientated, like Sam Altman, Larry Summers or Microsoft.
Even compared to Microsoft there will be companies who move faster, further, with less introspective ethical concern for the rest of us than even the inventor of Microsoft Teams.
The trouble at OpenAI caused by safety foresight or ethical concerns will serve as a major warning to all other companies not to fall into the same trap, and to just get on with the innovation and money making.
As a start-up, your investors only care about ethics up to a point. It is indeed a sad inditement of Western civilisation that the moment our virtue signalling endangers an investment or prestigious position, all the ethics seemingly go out of the window and the u-turns begin.
I deeply regret my participation in the board's actions. I never intended to harm OpenAI. I love everything we've built together and I will do everything I can to reunite the company.
— Ilya Sutskever (@ilyasut) November 20, 2023
Sometimes the dangers of a technology or risky science isn’t that the whistle blowers refuse to speak out or that power struggles happen to decide who profits – it is that it exists at all – and is inevitable.