“Nothing computers have done so far can compete with our brains in terms of originality or flexibility. They are not poisoned to outwit us, and perhaps striving to make them think more like us is the wrong approach. Instead, machines will always be most useful to us precisely because they think in very powerful but distinctly unhumanlike ways.” — Lord John Browne, Make, Think, Imagine: Engineering the Future of Civilization (2019)
“It’s literally making stuff up and it has no idea what it’s making up, therefore it’s a bullshit engine. It’s a bullshit engine in that it makes stuff up which has no semantic content, grounding or causality in it. And it’s bullshit because its only goal is to be plausible, just like somebody who tries to bullshit you at a party that they know all about fusion physics or something like that and then they don’t. They’re just bullshitting and they’re just trying to sound convincing.” — David McQuillan, Resisting AI: An Anti-Fascist Approach to Artificial Intelligence, on the podcast Tech Won’t Save Us.
*****
Among the things I never thought I’d see is Sam Altman before a Senate committee.
Sam is a friend and I once advised him in his embryonic campaign for governor. He wisely concluded he couldn’t beat Gavin Newsom, who is, after all, an adopted Getty. Still it was terribly interesting to see him chat about California politics with Dominic Cummings.
I wish Sam well in life and of the tech people I find him to be one of the least obnoxious of the tech people. He is warm and friendly and gentlemanly. All that said, I think it’s a wise idea for regulators in the US or UK to ban generative AI altogether.
Thank God the British are coming!
It’s very cool and long overdue that our British friends are riding to the rescue.
Just as AUKUS requires new submarines so too does it require sense-making about complicated topics. President Biden dropped in on the AI meeting at the White House the other day.
The truth of the matter is that Microsoft needs the Open AI stuff to work for its future.
Microsoft’s deal with Activision is dead — just as we predicted — and Microsoft’s cloud is way too compromised to be very valuable going forward. This is why it continues to lose contracts to Google Cloud and Amazon Web Services. Microsoft has become an Israeli and Chinese company as my friend Jeremy Rothe-Kushel has pointed out.
Here’s one frame:
Effective Altruists (AKA Likud cult), compromised Microsoft throwing cash at a nonsense project, and building their sexual blackmail networks...
This is not as wrong as people want it to be.
Are we really not going to talk about how the Open AI chief scientist moved from Russia to Israel and then to Canada and the U.S.?
As cofounder of Clearview.AI — we’re suing over this and I’m winning — I’ve been asked my opinion about what AUKUS ought to do when it comes to reining in AI’s excesses. Yes, I’m very pleased to see that this Substack is read by those close to the FTC and to the UK government.
Let’s understand first how AI works. AI has become the “fake because” which lets companies do whatever they wanted. They don’t want you to know they are firing people because of lack of liquidity, sales, and Chinese cash.
Nor do they want you poking around at the climate implications of giant server farms turned toward bullshit engines.
Speaking in generalities I agree with the sentiments expressed by FTC’s Lina Khan in the New York Times. AI can fuel market dominance and collusion by large firms, increase fraud and discrimination, she says.
Here are some of my thoughts:
We need a real security review of the people working on Open AI. I share Sinews of War’s concern about the CTO, especially. I think we should be asking if these people are spies.
Don’t let the neurotics set the policy. Trust the very humaneness and cultural differences of our two great countries. Broadly speaking the cultural differences between America and Britain can be found in their comedies. In America, a lovable group of slackers rise to the occasion. In Britain the comedy comes from a man struggling against the natural order of things.
We should follow the naturalists. Let the British Museum be your guide. AI needs to be about collecting, processing, organizing, supporting the natural world. Focus also on Deepmind’s applications for the natural world. Focus on AI’s uses to build geno-pheno matches.
Currently Open AI is going in the wrong direction for a number of reasons:
To crack down on misinformation we need to “know your customer” laws for the Internet. We should ban pseudonymous accounts, especially if they are running ops on the public. This effort will drastically cut down on ad fraud, which will, in turn, put downward pressure on the already distressed ad markets.
Microsoft’s cloud is leaky. You should see generally Nicole Perlroth’s This Is How They Tell Me the World Ends: The Cyberweapons Arms Race which goes into Microsoft’s leaky cloud. She doesn’t quite go there with the Israeli, Chinese, and Russian penetration. My friend Jeremy Rothe-Kushel points out the problem of JEDI and its national security implications on his YouTube channel.
Microsoft needs the regulation. Activision is a public health and national security risk for reasons we’ve already laid out. The Brits were right to stop the merger.
Open AI does have serious climate implications — just as crypto did before it. President Biden recently proposed a tax on cryptocurrency and a similar tax on generative AI should be deployed as well.
Open AI disguises itself as a solution to problems that require complex thought. As of now, it cannot and should not be used to diagnose anything healthcare related or provide insight into any legal argument.
Here’s the history:
When OpenAI provided Microsoft with the details of the scale it wanted to compute. Microsoft realized that it needed to give OpenAI access to its cloud computing services for an extended period of time.
So, Microsoft began stringing together tens of thousands of A100 graphics chips from NVIDIA that have the large computing power and are frequently used for developing AI models. However, it was not just a matter of connecting the chips together.
When training AI models, work is divided between the thousands of processing units that fire up at once, requiring meticulous power management. Beyond that, the processors also need to talk to one another to share the work done for which Microsoft had to write special software.
*****
As I was putting this piece to bed a friend wrote to me about a message he had received. I checked it out a little but I don’t speak Albanian. It looks like it checks out.
I can speak to Silicon Valley being anything but a meritocracy and I think fair minded people would note that.
Mira Murati is a fraud and it needs to be exposed before her rise as a girl boss goes too far
Her family member, Hekuran Murati, is Minister of Finance of Kosovo Albania
(https://en.wikipedia.org/wiki/Hekuran_Murati)
https://twitter.com/HekuranMurati/following
He also is an investor in Tesla.
Many people on Twitter are citing that she worked at Goldman Sachs and Tesla.
Her family had direct connections. Also look up Valon Murati. Even her being accepted into the international diplomacy program she had family connections.
Many young people on TikTok were rightfully confused how with someone with less than 1 year of experience was a Senior PM at Tesla
This is how. But the biggest question is why is it being funded by the Albanians? Albanians are known to the U.S. to be arms and drug dealers.The Albanian gov't is corrupt per their citizens' evaluation.
https://www.aa.com.tr/en/europe/albania-to-launch-2-satellites-into-space-in-2022/2397950
It was also terrible for Sam Altman to tweet about meritocracy and then promote Mira who is there on family connections.
https://euronews.al/en/sinaloa-cartel-allied-with-the-albanian-mafia-the-mexican-media-reveals-list-of-names/
You may find it hard to believe but I am a woman who actually is a programmer/developer. I find it upsetting that she is getting credit as the developer when she never even wrote a line of code. She was a mechanical engineer. If you go on ChatGPT is says who actually developed it and she is not there. She is a figurehead but she didn't write any code.
This is why the history months are a joke, they will uplift people from 50 years ago or people who aren't even even doing the work as "women in tech"
Anyone in tech finds her rise suspect because we know how jobs in tech work. a mechanical engineer with less than one year experience wouldn't become a senior PM. It's only people not in tech and enthusiasts who seem impressed with her background. Her background of interning at Goldman (which she got on connections not GPA like everyone else) then working of a tech company is very common. Look on LinkedIn. So you were right to ask why her for CTO since she came out of nowhere? She is not a Marissa Mayer. Her rise didn't make sense when there are more attractive women across the globe who have PhDs in Computer Science and AI.
https://openai.com/blog/chatgpt
Her name is not even listed on the contributors so I don't know why the news cycles chose to run with that she is the brains behind anything and say she is the creator. It's an insult to the actual woman software engineers in tech.
Also that one photo of her that is circling around is very extremely filtered, touched up, blurred and Photoshopped but the bar for what a for an attractive woman is very low in tech.
The misinformation needs to stop!
This is mighty curious.
Is Murati a spy? Is Ilya Sutskever — the Russian Israeli-Canadian Chief Scientist of OpenAI— a spy?
Is this a continuation of the same Israeli op that saw infiltration of spies into major tech companies? You know, the whole one organized by Benjamin Netanyahu… The one he bragged about incessantly?
Makes you wonder…
Our friend Sinews of War notes the following details:
Having looked into her education, there aren't any glaring red flags. Pearson college, Colby college/Dartmouth. Looks like UWC scholarship.
Then she shows up at a French aerospace company then Tesla. I'm leaning towards her getting corrupted at Tesla, but she could have been "got" at Dartmouth and put into Zodiac, which Safran — 737 MAX-linked company — was targeting.
We know her coworker Shivan had Musks IVF twins, so what did she do?
Perfectly normal to be invited to Chisraeli billionaire Barry Diller’s “wife” Diane von Furstenberg’s girl boss event.
OK. Sure. I believe this. Definitely.
****
The way I think about the difference here is generative vs. revelative, (or as we yanks say, “revelatory”).
Or put another way: bullshit engines against search engines. We could build search engines for the natural world if we wanted to.
A search engine isn’t useful if the information within it isn’t true, and all AI is wholly dependent on the quality of the information going in.
Open AI — and other bullshit engines — as they currently exist will ultimately make using the Internet a shittier experience as if it could get any worse.
Regulators: You have a moment to reclaim the deep state, deep tech roots of the Internet.
Seize it.
Repeat after me: We need you to be clever, my dear British friends, so that we can be authentic.