AI & Berklee

[ghmerrill]

All this talk about what Musk "does" in AI ... I don't think he is doing any of that work.

It would not have taken much effort with a search engine (avoiding an AI if you feel it may warp findings) to learn that this is incorrect -- depending on what the scope of "any of that work" may mean. I've certainly seen examples where, at a particular level of a technical area, Musk is out of his depth and takes a position that is wrong or incoherent -- which is a clear indication that he doesn't know everything about everything (though he may on occasion want to pretend that he does). History (including the history of science) is full of examples of highly competent people who have made fools of themselves at one time or another in this way -- or simply made mistakes. Again, my original point was that Musk's thinking on dangers of AI is both deeper and broader than others (particularly Altman). I don't need to idolize Musk or endorse everything that he does or says in order to have that perspective. I'll judge Musk's views on those dangers (and how to approach and address them) based on their own merits.

I'm going to bail out on this now. If people feel that Musk is evil, stupid, a moral reprobate, a danger to civilization, technically inept and/or ignorant of the various aspects and potential effects of AI -- and therefore his views on it shouldn't be considered -- then they certainly are free not to look at what he has to say (and what he's been doing) in these areas. But I think that would be a serious mistake.

[harrisonreed]

All this talk about what Musk "does" in AI ... I don't think he is doing any of that work.

It would not have taken much effort with a search engine (avoiding an AI if you feel it may warp findings) to learn that this is incorrect -- depending on what the scope of "any of that work" may mean. I've certainly seen examples where, at a particular level of a technical area, Musk is out of his depth and takes a position that is wrong or incoherent -- which is a clear indication that he doesn't know everything about everything (though he may on occasion want to pretend that he does).

So I'm not attacking you here. I just think that Musk is being represented (by a lot of people) as some kind of Tony Stark genius who created all the tech he is associated with. But that's not really what he does. He has resources, conducts his version of talent management, and sets goals for his teams that are generally overly ambitious.

He did not "invent" PayPal, but that's what made him rich after he acquired it. He doesn't do any of the AI coding or actual "work" on AI, he just uses his vast resources to set the conditions for bright people to do that work. He didn't invent the electric cars that Tesla sells, and in fact has hindered that company in many many ways (square cyber truck steering wheels...??? Alienating his major customer base through political action...?!?).

He is not a rocket scientist -- he claims he is because he read books .... but it's really Tom Mueller who is leading the engineering team and building the unbelievable engines and tech that goes into those things. Musk just creates goals and timelines that are impossible or highly impractical to reach, and wants things like "pointier rockets" and "Big ********** Rockets".

Musk is not doing the work, but he is hiring seriously bright people and then setting goals for them, some of which are overly ambitious but generate results (I want a 100% reusable rocket in 1 year!) and others that are just stupid (I want to make a giant ****-shaped rocket and make it extra pointy and call it the Big ********* Rocket and make sure the first launch is on 20 April because that is sooo hilarious), and others that are downright certifiable (he is pro eugenics and has offered to be a surrogate father (IVF) with his female friends and acquaintances) ..... (.... Colonizing Mars as an alternative to planet Earth .... ). It's pretty bad that the whole reason he created SpaceX was so that he could try to convince 1M people to fly on a spaceship to Mars and colonize it. And I really do like SpaceX. But that's something a crazy person who uses gobs of ketamine would think is a good idea. Mars has like 38% of Earth's gravity, it's highly irradiated, it has very little water -- terrestrial life cannot survive there even on a spaceship for too long. It is a terrible idea to build a city there.

[bwilliams]

Jesus christ. I'm done here.

[robcat2075]

Your job will be done by AI.

Really done.

starting at 3:17

[JohnL]

Musk is not doing the work, but he is hiring seriously bright people and then setting goals for them, some of which are overly ambitious but generate results...

I think maybe part of his skill set is a knack for convincing those very smart people that those overly ambitious goals are actually achievable.

Sometimes he's awesome and sometimes he's awful. Sometimes he's both at the same time. I do wonder how history will judge him.

[AndrewMeronek]

I ran across this related article, apparently from May of this year (2026):

https://time.com/article/2026/03/26/ai- ... the-tide-/

In late February, the British singer-songwriter Benedict Cork posted a snippet of himself playing a soulful new song called “Something Kinda Strange.” The clip quickly racked up more than 100,000 plays on TikTok, with fans clamoring for the full song.

A few days later, Cork started receiving messages about the song’s release—which was odd, because he hadn’t finished it yet. Eventually, someone sent him the song, which was on streaming platforms under the name Eduardo Arguelles. Not only did this rendition have full instrumentation and backing vocals, but it also had a new second verse and bridge, complete with lyrics that Cork begrudgingly wished he had written—including “Something kinda dangerous in the way you say my name.”

This version of his song had been created with AI. Someone had seen Cork’s snippet, and, trying to capitalize on its virality, ran it through an AI music generator and posted the result in order to generate streams. “At first, I found it really funny. Then I was impressed by how amazing the technology is,” Cork says. “And then I became a little more angry.”

My take: that's using Benedict's music without permission, and also he should be part of the revenue stream which this AI slop got from streaming. That is my argument here: AI companies like Suno need to be required to compensate the artists used to train their AI. This article doesn't actually make this argument, instead bemoaning the theft without thinking further.

Think of what happened with Napster: people first only thought about the theft. It took too long for people to realize that the streaming/file sharing technology was here to stay and the question really was about who gets paid.

This is an essential question for every musician here.

[Digidog]

What baffles me in the debate as a whole, is that nobody questions the fact that one single societal phenomenon gets to consume such enormous amounts of resources from the rest of the society. If anyone grunts about rapidly increasing costs for electricity, or shortage of power for the system as a whole, the companies that venture in AI are to blame for the major part of that. How politicians and citizens still not oppose that shameless appropriation of energy and insatiable demand for more, is - as said - baffling.

The second thing that irks me, is that those calculating programs still are regarded as intelligent; they are not and will by default never be. They can be immensely powerful at calculating, but they can never ever be intelligent.

If it were up to me, I’d pull the plug on the AI companies - or at the very least make them pay proper prices on their resource consumption and an adequate tax on their businesses. As it is now, they are loose, unbridled, misguided fools that promote a tool with no over all benefit whatsoever for either humanity, nor the planet.

I will avoid using any of those programs long as I possibly can.

[harrisonreed]

What baffles me in the debate as a whole, is that nobody questions the fact that one single societal phenomenon gets to consume such enormous amounts of resources from the rest of the society. If anyone grunts about rapidly increasing costs for electricity, or shortage of power for the system as a whole, the companies that venture in AI are to blame for the major part of that. How politicians and citizens still not oppose that shameless appropriation of energy and insatiable demand for more, is - as said - baffling.

Yes! We do not have enough clean energy yet for this to make sense. The vast majority of that energy is being wasted on making deep fakes of Stephen Hawking winning WWE wrestling matches, where it could be being used instead on models trained to look for new medicines and stronger alloys.

The second thing that irks me, is that those calculating programs still are regarded as intelligent; they are not and will by default never be. They can be immensely powerful at calculating, but they can never ever be intelligent.

Never say never. Also, we already have Boston Dynamics robo-dogs that can go into dangerous buildings, map the area autonomously, find people who need to be rescued autonomously, operate machinery in areas that are hazardous to humans, etc etc. You just give it the command to go on and check for survivors. How is that not intelligent? AI and especially AGI (it's 3 years away) are just alien forms of intelligence. Hell, bees are intelligent, it turns out. They work in teams to solve puzzles and get sweet treats. We used to think that bees were basically robots. But they aren't.

If it were up to me, I’d pull the plug on the AI companies - or at the very least make them pay proper prices on their resource consumption and an adequate tax on their businesses. As it is now, they are loose, unbridled, misguided fools that promote a tool with no over all benefit whatsoever for either humanity, nor the planet.

I will avoid using any of programs long as I possibly can.

I agree, except I do think AI research should continue but just only be available to government and research institutions -- I've been reading about the numerous ways that these AI companies are worming out of paying tax or fair prices for the power they consume. It's pretty bad.


Then again anyone owning an S&P500 index fund is complicit. Guilty as charged...

[robcat2075]

If it were up to me, I’d pull the plug on the AI companies...

There's a large constellation of AI companies (45% of the S&P 500 by one measure) with stock values driven up, not by any current profitable results, but by mere presumptions... promises... that the AI stuff will be insanely profitable in the future.

Approximate Stock Price to Earning ratios for several notable AI-related companies

Upstart 60
AMD 135
Palantir 145
ARM 250
Tesla 400
Crowdstrike -700 (that's a negative number)

A normal P/E ratio is maybe 20


If you pull the plug now, the rich people who seeded these companies will have nothing to show for all their money venturing.

You have to wait until they have exited their investments so that the cost the AI Economy Collapse may be borne by everyone else.

[GabrielRice]

https://futurism.com/artificial-intelli ... r-booed-ai

Listen to the students who are living it! They know and they don't want it!

[harrisonreed]

The AI economy collapse might happen but it'll be different than CEOs exiting the bubble. From what I can glean it will more likely take the form of a complete economic paradigm shift.

Even LLMs are capable of doing a week's worth of a white collar employee's work in a few hours.

There is even a GitHub project, GitHub squad, a toy to be sure, that simulates a software development team. You prompt it with the type of program you want and it goes in an actually simulates board meetings, codes beta versions, beta tests, conducts progress briefs, and publishes a final product (you approve or reprompt it's work). You can go in and see where the team decides to hold a special meeting to debate removing or adding a feature after the initial beta test results (unprompted). And that's a free toy available on github.

The economy will change if AI continues to progress, and we might all be left in the dust. I believe the ability to perform real work doubles every five months or so. Corporations that can run these modeled teams won't need humans to do any of the work.

[Digidog]

The second thing that irks me, is that those calculating programs still are regarded as intelligent; they are not and will by default never be. They can be immensely powerful at calculating, but they can never ever be intelligent.

Never say never. Also, we already have Boston Dynamics robo-dogs that can go into dangerous buildings, map the area autonomously, find people who need to be rescued autonomously, operate machinery in areas that are hazardous to humans, etc etc. You just give it the command to go on and check for survivors. How is that not intelligent? AI and especially AGI (it's 3 years away) are just alien forms of intelligence. Hell, bees are intelligent, it turns out. They work in teams to solve puzzles and get sweet treats. We used to think that bees were basically robots. But they aren't.

Intelligence is the mutual interaction between a sensory-motoric input, calculative ability and physical conditions that permit or force physical adaption and development, both of the calculating functions and/or the physical traits. Those robo-dogs can’t procreate, nor feed themselves (they can go to a charging station, but that’s not equivalent to feeding), nor develop anything they were not equipped with from the start (their calculations can be better, but their calculating functions do not develop), nor can their sensory systems develop their calculating and decduction skills nor evolve from what they had from set-up. So they are calculating, but not intelligent.

However much a company will develop and cram in cumulatively stacking algorithms and sensational back-data adaptations in a machine, a machine will never have the sensory-mototric input and adaptability of a living organism, nor will it be subject to those processes that necessitate or force it to adapt, adjust and develop independently. There is a logical-mathematical law that says that it is impossible to include an outer axiom into a deduction that is depending on that very axiom. In other words: If you deduct that there is no difference between a rock and your mom because they both cannot fly on their own accord, you cannot include what makes flying possible since you thereby expand the terms for your comparison and thus your conclusion to factors that make your reasoning void. Regarding our Universe, and the conditions for life and matter: You cannot lift yourself by yourself, because if you could, you would include forces (factors) that would render your position [EDIT: i.e. the logical conclusion of what your existence is] in the Universe [EDIT: a deterministic Universe] void.

To be clear: No human artificial creation can be guided and developed by factors that govern the creations of life, since we cannot create a framework within an overarching [EDIT: a deterministic] framework that includes the causes and effects of that framework. The closest we have ever come to that, is the nuclear energy and the nuclear bombs, and those phenomenons are - as we now know - deterring to their cause and effects. The DNA/RNA manipulation isn’t even close.

Then again anyone owning an S&P500 index fund is complicit. Guilty as charged...

[harrisonreed]

Intelligence is the mutual interaction between a sensory-motoric input, calculative ability and physical conditions that permit or force physical adaption and development, both of the calculating functions and/or the physical traits.

The robo dogs go into a building and then behave based on unknown conditions that are present in the building using their cameras. So they fit this description. "Development" could be considered as a condition of this definition that is met because they adapt their behavior to meet the unknown conditions inside as those conditions become known.

Also, my favorite quote from Dune kind of fits this:

"Think you of the fact that a deaf person cannot hear. Then, what deafness may we not all possess? What senses do we lack that we cannot see and hear another world all around us?".

AI just has different sensory inputs from animals.

Those robo-dogs can’t procreate, nor feed themselves (they can go to a charging station, but that’s not equivalent to feeding), nor develop anything they were not equipped with from the start (their calculations can be better, but their calculating functions do not develop), nor can their sensory systems develop their calculating and decduction skills nor evolve from what they had from set-up. So they are calculating, but not intelligent.

This feels like the definition of life, not intelligence. Slime mold can do some of this, and solve mazes, but we don't think of it as intelligent.

From what I've been reading, AI is already writing over 90% of the code to improve successive versions of itself, and we are very close to recursive self-improvement.

However much a company will develop and cram in cumulatively stacking algorithms and sensational back-data adaptations in a machine, a machine will never have the sensory-mototric input and adaptability of a living organism,

Recent videos I've seen produced by Anton Petrov (a popular science commentator) suggest that the definition of what constitutes life and an organism is being blurred. They recently discovered a bacteria that devolved into a virus. Is it no longer, then, an organism? I think we need to realize that an AGI will be essentially an alien intelligence. So trying to define it within the constraints of "life" will probably set us up for some nasty surprises.

nor will it be subject to those processes that necessitate or force it to adapt, adjust and develop independently.

https://thenextweb.com/news/anthropics- ... release-it

Won't it?

There is a logical-mathematical law that says that it is impossible to include an outer axiom into a deduction that is depending on that very axiom. In other words: If you deduct that there is no difference between a rock and your mom because they both cannot fly on their own accord, you cannot include what makes flying possible since you thereby expand the terms for your comparison and thus your conclusion to factors that make your reasoning void. Regarding our Universe, and the conditions for life and matter: You cannot lift yourself by yourself, because if you could, you would include forces (factors) that would render your position in the Universe void.

I sort of get where you are coming from here, but AI doesn't exist in a vacuum, either. Humanity has certainly moved itself. An AGI with access to compute and data (cameras, feedback from humans, feedback from drones it controls, etc) could do the same. You can't pull yourself up without anything, but you can pull yourself up on a pull-up bar. AI can do the same with compute and access to experimental or experiential data.

You can't try to dictate what alien life might be like using the rules that govern life on earth. When an alien shows up with two legs, two arms, and two eyes in a movie, with a ray gun ... that's so unrealistic and unlikely. Alien life and intelligence will likely be unrecognizable as such to humans.

An AGI will be the same -- completely alien. You can't use the definition of life to define artificial intelligence.