7 key takeaways from Elon Musk's latest conversation with Lex Fridman about the future
Last week Lex Fridman posted a long conversation with Elon Musk as part of a mammoth eight and a half hour podcast. These are seven takeaways that are worth paying attention to.
Elon Musk talks a lot. And not all of what he says is worthy of writing about. But I think his latest interview with computer scientist and podcaster Lex Fridman, which aired August 2 as part of a mammoth eight and a half hour podcast (yes, you read that right), is.
There are a couple of reasons for this. First, whatever you think about Elon Musk and his ideas, it’s hard to deny that he has an outsized influence on how people think about and approach the future. And so, despite all of the controversy that he attracts, I’d be remiss if I ignored where his thinking is going in my work around advanced technology transitions and the future of being human.
And second, he has a great rapport with Lex Fridman (this is his fifth time on the podcast), which means that in the conversation just released there was more candor and less posturing than is sometimes the case.
Of course, Elon tends to be a Rorschach test for many people, with how they interpret what he says and does being as much a reflection of what they think about him as what they see and hear — and I doubt this interview will be any different.
Nevertheless, where Musk’s head is at and how this potentially impacts emerging technology transitions — as well as what it might mean to be human in a technologically advanced future — is deeply relevant to my work and to my students.
And so I was interested to hear what he had to say in his conversation with Lex.
The podcast episode as a whole focuses on the work of Elon Musk’s company Neuralink, and includes interviews with Neuralink scientists and their first implant subject, Noland Arbaugh.
At eight and a half hours it’s crazy-long, but worth exploring for insights into Neuralink’s work (and fortunately there are time/chapter markers that allow you to enter the podcast at different points).
The interview with Elon is, in contrast, a mere one and a half hours long. But as even this is probably too long for most people to digest, I thought it worth pulling out seven aspects of it that stood out to me — including relevant clips from the conversation and links to the original video (where you can also watch with subtitles if necessary).
These are:
1. Altered Humans
The primary focus of Neuralink at the moment is to address severe medical conditions with advanced brain computer interfaces (BCIs). However, the conversation with Elon almost immediately dived into human enhancement, and the idea of BCIs substantially extend current human abilities:
(View on the original podcast)
Elon argues that Neuralink is heading toward a technology that massively increases data transmission rates between a human brain and a computer to the extent that it will vastly exceed what is currently possible through writing and reading or speaking and listening — and that this could lead to a future of being human that is very different from the present of being human.
There are, not surprisingly, some major issues with this way of thinking — including the assumption that the speed of data transfer is synonymous with the speed of understanding transfer.
But putting that aside, Elon goes on to speculating what this might mean if people equipped with Neuralink devices could communicate and engage with computers at speeds that far surpass what is currently possible.
The speculative result — and part of Elon’s vision of the future — AI-human symbiosis:
(View on the original podcast)
I suspect that this will have many neuroscientists yelling at the screen in frustration — and for good reason, as there are a mountain of simplifications being made here around how the brain works and what is possible with BCIs.
But it’s worth paying attention to nevertheless, as Neuralink is actively building the foundations and technologies to make BCIs highly accessible in ways that I don’t think any other companies in this domain are. And it’s being driven by a vision of the future where millions of people and more can engage with machines, AI, and each other, in ways that currently lie beyond our understanding.
This whole endeavor might, of course, collapse. But it might not. Either way, the idea of creating altered humans is one that’s probably worth paying attention to.
2. Better Than Human
Embedded in this idea of altered humans is the concept of using BCIs to exceed what are considered to be biologically constrained “normal” human abilities.
Skipping forward a few minutes in the conversation, Lex comes back to the question of communicating with computers using BCIs. Elon initially pulls things back to Neuralink’s current focus on addressing disabilities such as motor control and sight. But he rapidly jumps forward to thinking about how BCI technologies that “fix” deficiencies can also be used to “improve” people:
(View on the original podcast)
For anyone who’s watched the movie Transcendence or read the book Films from the Future, this has a very “Will Caster” vibe to it — a character in the film who merges with an AI and uses its resulting superhuman intelligence to “fix” people — including bringing sight to the blind in a scene that is both Biblical and very much in line with one of Neuralink’s current ambitions.
Again, I think that Elon is being both naive and hyper-speculative here. But I wouldn’t dismiss the momentum he’s building around using medical interventions to give people abilities that go far beyond beyond what’s considered normal.
And from the perspective of the future of being human, the aspirations driving Neuralink and the technological steps that are currently being taken should absolutely be part of a much wider conversation around where we collectively want to go as a species.
3. Digitally Altered States of Consciousness
Lex’s conversation with Elon was, at times, meandering (although this is also part of it’s serendipitous charm). But on one of those meanders something came up that I’ve long found intriguing: The possibility of computer-induced altered states of consciousness.
We join the conversation below as Lex has just described the experience of imbibing a large dose of the psychoactive drink ayahuasca deep in the Amazon jungle. It’s a great exchange as it shows the genuine rapport between Lex and Elon. But more interestingly, it leads to Lex asking “can I get that same experience in a Neuralink?”
(View on the original podcast)
Elon’s answer is “probably, yeah.” He goes on to explain his perspective that the whole of a person’s perception of reality comes down to electrical systems in the brain which means that, in principle, any experience is replicable.
I’m not sure I fully buy into the “electrical signals are everything” model as there are layers of electrochemical complexity in ho our brains work — even if you discount ideas around quantum effects being involved in consciousness. Despite this though, the idea that a Neuralink implant is simply a generalized input/output device is an important one.
If Neuralink and other BCI companies are developing general purpose read-write interfaces that are designed to modulate our perceptions, it opens up a whole raft of possibilities — including psychoactive experiences and altered states of consciousness.
There are, naturally, some really big questions here — including how localized electrical signals stack up against the more distributed modulations of brain chemistry associated with chemical psychoactive substances. But just the fact that people are already beginning to think about how BCIs could lead to perception-altering, mind-altering, and life-altering experiences, should be food enough for thought for anyone interested in tech and the future of being human.
4. Safe Artificial General Intelligence
This is a topic that took me by surprise — the idea that Neuralink implants may be a one component of ensuring the safety of super-powerful artificial general intelligence, or AGI — but it’s an intriguing one.
This may sound a little crazy, but Elon’s argument here made me smile. It’s based on the idea that we’ll find it incredibly hard to ensure alignment between human and AI values in the future if we can’t communicate with AI effectively. And to do this, we’ll have to learn how to communicate far faster than we can achieve through our biological fingers and voice alone.
The analogy Elon draws in the conversation is with trying to value-align with a plant — a little difficult if the plant is trying to communicate with us so slowly we don’t even realize what it’s doing.
Here’s the clip:
(View on the original podcast)
This is nudging things well into sci-fi territory. But also, again, the important thing here is not how wild the ideas are, but how much influence they have over how the future unfolds.
Plus, I rather like the analogy between us engaging with plans, and AGI engaging with us!
5. Training Data for AI
Shortly after discussing BCIs and AGI alignment, the topic turned to training data for AI systems — something of a topical issue as fears mount that we’re hitting the limit of human-generated training data.
Apart from a throwaway comment about most AI companies having already used Twitter data for training (and yes, Elon called it Twitter), what stood out in this part of the conversation — and should make people sit up — was Elon’s comments about Tesla cars and robots gathering massive amounts of valuable new training data — in his words generating “useful data from reality”.
In the case of Tesla cars (and trucks of course), these are mobile and ubiquitous devices that are bristling with cameras which capture the world around them in fine detail. And the more Teslas there are on the road, the more real-world data is collected that captures every conceivable aspect of the physical and human world we inhabit.
Using devices that are — or may be — part of our everyday lives to collect training data for AIs raises a mountain of complex questions around data privacy and security. There’d be outrage if the cameras and microphones on our phones were permanently on and being used to train AIs (although — despite this being illegal in some places — I suspect that some people believe it’s already happening). But I’m not sure anyone’s asking serious questions about cars and future humanoid robots.
Given the incredible amounts of data that must be being collected on everything — everything — that every Tesla vehicles see, you can understand the attraction of using these data for training purposes.
But this pales into insignificance compared to Elon’s vision for the data collected by Tesla’s humanoid robots, if and when they become ubiquitous.
These will be, if his aspirations are realized, an unprecedented source of training data on how the world works — from the physics of cause and effect, to (presumably) everything that the robots see, hear, touch, and experience.
And of course, while Tesla’s have to stick to the road (typically), robots do not.
Definitely worth remembering as you consider your Optimus personal companion or home help in the future …
(View on the original podcast)
6. A Billion Plus Humanoid Robots Per Year?
Leading on from the conversation around AI training data, Lex and Elon got onto talking about the future of humanoid robots — Tesla’s Optimus robot in particular.
Remembering that Elon is not known to shy away from overpromising, he speculated that we could be looking at a point in time when over a billion humanoid robots per year are being produced.
This is an audacious vision — and one that would be transformative if it came about.
Whether it will will, of course, depends on more than the technology itself. There’s consumer interest, consumer acceptance and perceived value, the economics of production, supply and demand — and a highly uncertain regulatory and geopolitical landscape. All of this means that the success of technologies like this depend on more than engineering smarts (which, as an aside, is why we need to be training a new generation of students who are well versed in transdisciplinary approaches to supporting the emergence of societally beneficial technologies)..
But again, what is important here is the vision that Elon is expressing, coupled with very real investments in making it a reality — rather than simply dismissing his ideas as ridiculous.
(View on the original podcast)
7. Technology Policy
The final point I want to highlight in the conversation is one that’s got quite a lot of play on social media. This is where Elon begins to speculate about how government needs to be fixed to better-support Musk-style human flourishing through technology innovation:
(View on the original podcast)
The idea here is that, over time, expanding government regulation leads to a “hardening of the arteries” of an economy and stifles progress. This isn’t a new idea. But it is one that has growing political cachet in some tech quarters — and it isn’t wholly wrong.
Whether Elon has the political science or governance understanding to help a future government navigate the tension between transformative technologies and ensuring life, liberty, and the pursuit of happiness for all people, is doubtful. But he does have spades of influence.
My read is that the conversation here on science and technology policy is just two tech dudes shooting the breeze. But if we do get the a point in the near future where Elon and others with a similar mindset are putting their thumbs on the scales of technology policy and governance, I hope that there’s an informed counterbalance.
Final Thoughts.
At the time of writing, this latest Lex/Elon podcast has been viewed over 1.5 million times (averaging over half a million views a day). Even accounting for the vast majority of viewers who wouldn’t have stuck out the full eight and a half hours, that’s a lot of reach.
This in itself makes understanding how Elon’s thinking about the future, and how this in turn could influence the course of a bunch of transformative technologies, important.
But — and I suspect this will divide my readership — I also found the conversation enjoyable to listen to.
Putting aside the very real controversies around Elon — from his use of X as a platform for political and social influence, to some of his driving ideas around humanity and the future — I get where he’s coming from. I enjoy his flights of speculation and his willingness to roll up his sleeves and try to make them reality. And I suspect that, when talking about the future of being human, we’d get on well.
This makes his more egregious actions all the more frustrating. But it does mean that, for all the controversy, I think he’s worth listening to.
Even if it’s just to be prepared for where he and others aspire to take the future, whether we’re behind him or not.
Very useful breakdown, thank you! I think I'm interested enough to listen, but I have these other things like a job and family to think about, so we'll see! It reminds me that I forgot to share my OpenAI story with you. Up to part 3 already. https://shonistar.substack.com/p/what-totally-went-down-at-openai
Thanks for wading through the whole tape and providing this quick gloss. Elon is a truly big thinker and this was my first intro to his ideas about neuralink.