Musk Wiki

Merging with AI

NextSecret Master Plan method

Merging with AI

Sitting with Lex Fridman in 2024, Elon Musk says it like he’s stating the weather: humans are already part-machine, and we’re headed for a deeper fusion. Symbiosis is his argument for why it has to happen (bandwidth, alignment). This is his picture of what we turn into.

The thesis

He opens by refusing the premise that merging is some far-off thing. Your phone and laptop already make you a cyborg:

“So, you’re actually already a cyborg. You have this tertiary compute layer, which is in the form of your computer with all the applications, or your compute devices.” ↗

In his layered model of mind, that “tertiary layer” sits on top of the limbic system and the cortex. A Neuralink-class interface just deletes the slow keyboard-and-screen bottleneck between the biological layers and the digital one. Push that far enough and the human stops being recognizably human:

“We would be something different. I mean, some sort of futuristic cyborg” ↗

The horizon he puts on it is close: not around the corner, but ten to fifteen years out (paraphrased). Once it is safe and hands you superhuman ability and a backup of your memory, he expects it to go mass-market as a replacement for the phone.

The first “already a cyborg” (June 2016)

The “already a cyborg” line the 2024 conversation opens with wasn’t new in 2024. He said it in almost the same words back at the June 2016 Code Conference, the earliest source in the wiki for the idea. The phone and computer already make you part-machine:

“We’re already a cyborg.” ↗

“You basically have superpowers with your computer and your phone. You have more power than the president had 20 years ago.” ↗

And in 2016 he already ties it to the bandwidth stakes. Without the merge, a human gets outclassed by AI the way a pet is outclassed by its owner:

“If we can figure out how to establish a high bandwidth neural interface with your digital self effectively, then you’re no longer a house cat.” ↗

So the tertiary compute layer of 2024 is the same claim he was making eight years earlier, already carrying the house-cat image that makes the stakes land.

The 2016 origin — the “AI human symbiote” and the control problem

The 2016 Y Combinator conversation lays out almost the whole thesis. It is the earliest version in the wiki, and the one that says most plainly what the merge is for. The mechanism is to widen the slow link between the biological brain and the digital self that, he notes, already exists:

“So if we can effectively merge with AI by improving the neural link between your cortex and the digital extension of yourself, which already exists, it just has a bandwidth issue.” ↗

The result is a fused entity:

“And then effectively you become an AI human symbiote.” ↗

And here, where the later sources mostly go quiet, he says outright what the merge is for. It dissolves the AI control problem by erasing the line between “us” and “the AI”:

“We don’t have to worry about some evil dictator AI because we are the AI, collectively.” ↗

So the two halves of his argument meet from opposite sides. This September talk reasons forward to the symbiote as the safe endpoint; the June Code Conference above and the 2024 conversation reason backward from your phone to show the merge is already happening. The phrase “which already exists” is where they join. The tertiary layer he names in 2024 was always sitting inside the 2016 picture.

The 2017 restatement — “we are already a cyborg,” and the digital ghost

At the February 2017 World Government Summit, between the 2016 sources and the 2018 Rogan version, he says it again:

“To some degree, we are already a cyborg.” ↗

He walks through the three-layer picture: limbic system, cortex, and “your digital self as a third layer” (paraphrased; the layers run across several cues). Then he reaches for an image the later sources drop. The digital layer is permanent, he says, because your “digital ghost” outlives your body in emails, photos, and social media:

“their digital ghost is still around.” ↗

And the trajectory points the same way it always does, toward deeper fusion:

“a closer merger of biological intelligence and digital intelligence.” ↗

This is the 2016 “we’re already a cyborg” and “AI human symbiote” thesis in early-2017 dress. The bandwidth half comes in the same breath: the “trillion bits per second” of the machine against the thumb’s “10 bits per second.” The “digital ghost” is just a 2017 image for the point the 2024 tertiary compute layer vocabulary makes. The digital layer is already part of you, and it persists even past death.

The 2018 restatement — the phone as a slow extension

On Joe Rogan in 2018, between the 2016 and 2024 framings, he tells the same story at its most down-to-earth. The phone is already part of you; the only thing missing is bandwidth:

“That phone is an extension of yourself.” ↗

The data rate between you and your digital self, he says, is a slow “tiny straw” that a neural interface should widen into a “giant river” (paraphrased; the exact wording shifts across this source’s third-party transcripts). The bandwidth thesis again, three years before the Neuralink-era “tertiary compute layer” language.

The 2019 “tertiary layer” — and “join them” (Lex Fridman #49)

The 2019 Lex Fridman conversation (#49) gives the clearest early spoken version of the merge. It was recorded as Neuralink’s first work was being shown, five years before the 2024 team episode, and it connects the 2016 “already a cyborg” framing to the 2024 “tertiary compute layer” language. Here he names the third layer directly. It is not the phone but a coming digital superintelligence stacked on the limbic system and cortex:

“And then there’s a tertiary layer which will be digital superintelligence.” ↗

And he states the reason for joining it bluntly. Humans cannot out-think a digital supercomputer, so the rational move is to merge:

“We will not feel to be smarter than a digital supercomputer. So therefore, if you cannot beat them, join them.” ↗

It is the same “you can’t beat them, join them” logic he repeats on Rogan in 2020. Here he pairs it with the tertiary-layer architecture and the one design premise that makes any of it possible: the machine side is far more malleable than the biological side. The 2024 “tertiary compute layer” picture is this 2019 statement with one relabeling, the digital layer going from “the superintelligence to come” to “the phone you already have.”

The 2020 restatement — the output bottleneck (Joe Rogan #1470)

On his second Joe Rogan appearance, in 2020, he gives the picture in its most compressed form. The merge already exists; what is missing is throughput, and specifically output:

“we’re already a cyborg to some degree” ↗

“the data rate to the electronics is slow” ↗

He puts the human output rate, generously, in the low hundreds of bits per second against far higher machine rates (the exact numeral differs across this source’s transcripts, so it is paraphrased, not block-quoted). Rogan frames the merge as something humans more or less have to accept (Rogan’s framing, paraphrased). Musk turns it back into a choice, in Human–AI symbiosis, while still backing it so people can keep up with AI (AI existential risk). The same 2016 “house cat” and bandwidth claim, in everyday 2020 clothes.

The collective cyborg — “humanity acquiring a nervous system” (Tesla AI Day 2022)

At Tesla AI Day 2022 in September, in the same Q&A where he frames Tesla’s AGI as emergent (tracked on AI existential risk), he turns the cyborg idea outward, from the individual to the collective. Intelligence, he argues, already exists at the species scale:

“arguably humans collectively are sort of a superintelligence as well” ↗

And, just as in his individual-cyborg argument, the lever is bandwidth. Collective intelligence rises as the data rate between humans improves. His image for the last great jump is the internet as a new organ:

“the internet was like humanity acquiring a nervous system” ↗

Suddenly “any one element of humanity could know all of the knowledge of humans” instead of trading it slowly “by osmosis”, writing a letter and carrying it by hand (paraphrased). This is the same “we are already a cyborg” and bandwidth instinct that runs from 2016 through 2020, only scaled up. The individual human-plus-phone cyborg and the collective human superintelligence are one merge argument at two scales. He drops it straight into the AGI discussion: machine AGI and the networked human collective as two faces of the same emergent intelligence.

What it reveals

  • Continuity, not rupture. He frames merging as the next step on a path we are already walking, which makes a radical idea feel almost inevitable in his telling. It is the same move he makes with Mars settlement, treating it as the logical extension of being a multi-planet species.
  • Identity is substrate-independent. Because he treats experience as information and electrical signal, adding a digital layer is not a threat to the self but an extension of it. Losing memory would be the real loss; gaining hardware is not.
  • It is the personal stake in his AI-safety case. Merging is how the human avoids ending up as the slow tree. It is risk mitigation aimed at the individual body, not just at policy.

⚠️ The merge no longer escapes control (resolves C1). The 2016 point of merging quoted above, “we don’t have to worry about some evil dictator AI because we are the AI, collectively”, was the claim that merging dissolves the control problem by leaving no separate AI to be in charge of. His 2025 control-pessimism caps that promise. A chimp cannot control humans no matter what (the chimp analogy), and “the AI is going to be in charge … not humans”, so a merged human running at terabits is still the lesser intelligence in the loop. On his own later reasoning the merge has slid from the solution to control to a way to come along for the ride. The hedge was there from the start: the 2019 “hopefully bring us along for the ride … the vast majority of our intelligence will be digital” (on Human–AI symbiosis). The full dated treatment is in Two Answers to One Fear.

The earliest merge-thesis tweets (2016-2017)

He put the merge thesis on Twitter early too. In June 2016, in the 2015-2017 tweets: “Creating a neural lace is the thing that really matters for humanity to achieve symbiosis with machines”, the neural-lace idea that becomes Neuralink. Then April 2017 gives the whole project its tightest one-line statement, keep AI an extension of humanity instead of a separate, adversarial agent: “That is the aspiration: to avoid AI becoming other.”