


Elon Musk stirred the pot during a recent interview with Bill Maher about whether AI will lead to ... [+]
They say that you only live once.
They also say that you need to make every moment count.
Suppose that you did live only once, but that you were able to live forever. Would you still need to make every moment count? Perhaps you would treat each moment as something that was readily tossed away. There would always be another moment up ahead awaiting you. An endless series of moments.
Why all this talk about a seemingly preposterous or assuredly outsized notion of being able to live forever?
Because Elon Musk reopened or shall we say awakened an ongoing debate in the field of Artificial Intelligence (AI) about how AI might just enable humans to live forever.
Yes, that’s right, the dreaded advent of a future involving AI that is an existential risk and wants to enslave or kill off humans is also speculated to be an AI that might enable humans to have immortality.
Well, to be frank, some say that it might not be immortality and instead just that humans can live a very long time. A lavishly extensive period of time. Possibly on the order of a thousand or thousands of years. I doubt that we would look askance at being able to live thousands of years and bemoan that it wasn’t the same as living forever. Probably best to be happy with thousands and not get finicky about the actual end of the line.
During a recent interview that Elon Musk had with Bill Maher (posted on social media), the topic of AI dangers was brought up amid concerns that we are rapidly heading toward those proverbial AI overlords. That seems to be the key fearmongering fad going on right now about AI. The emergence of generative AI such as ChatGPT has gotten the world on edge that maybe we are nearing sentient AI.
Just to be abundantly clear, we don’t have sentient AI today. I realize that might be a shocking revelation. Blaring headlines appear to suggest otherwise. Don’t fall for it.
What is good for the goose is also good for the gander, namely that the all-powerful AI that we are presumably slipping and sliding toward could also proffer benefits too. The question becomes whether we can corral the AI For Bad and keep it within reasonable bounds, meanwhile relishing the AI For Good that might provide an amazing array of positives for humankind.
In today’s column, I will take a look at the controversy associated with devising AI that can seemingly lead to human immortality or something close to it. You might be a bit surprised at the logic used to connect advances in AI with the chances of humans living forever. Hint: It probably isn’t due to what you think it is.
Into all of this comes a slew of AI Ethics and AI Law considerations.
There are ongoing efforts to imbue Ethical AI principles into the development and fielding of AI apps. A growing contingent of concerned and erstwhile AI ethicists are trying to ensure that efforts to devise and adopt AI takes into account a view of doing AI For Good and averting AI For Bad. Likewise, there are proposed new AI laws that are being bandied around as potential solutions to keep AI endeavors from going amok on human rights and the like. For my ongoing and extensive coverage of AI Ethics and AI Law, see the link here and the link here, just to name a few.
The development and promulgation of Ethical AI precepts are being pursued to hopefully prevent society from falling into a myriad of AI-inducing traps. For my coverage of the UN AI Ethics principles as devised and supported by nearly 200 countries via the efforts of UNESCO, see the link here. In a similar vein, new AI laws are being explored to try and keep AI on an even keel. One of the latest takes consists of a set of proposed AI Bill of Rights that the U.S. White House recently released to identify human rights in an age of AI, see the link here. It takes a village to keep AI and AI developers on a rightful path and deter the purposeful or accidental underhanded efforts that might undercut society.
I’ll be interweaving AI Ethics and AI Law related considerations into this discussion.
Getting From Mortality To Immortality
We are ready to further unpack this mind-bending matter.
Let’s reasonably assume that we are able to continue to make impressive advances in AI.
Some believe that we can have advanced AI that is ostensibly not sentient and yet nonetheless able to do incredible feats that seem to be akin to human thinking. Thus, one argument is that we don’t need to have AI attain sentience to garner all manner of vital benefits. In the same breath, this non-sentient AI can still be an endangerment. If we connect AI to our weapon systems or other crucial infrastructure, it is conceivable that AI will inadvertently go amok and harm humanity (see my discussion at the link here). This doesn’t need to be a result of the AI reaching consciousness and deciding that humans have to go.
A much less spectacular AI attainment can still produce existential risks for us all.
Envision that the non-sentient AI makes mistakes or potentially encounters so-called AI hallucinations (an increasingly accepted terminology that I loathe as it anthropomorphizes AI, see my discussion at the link here). Oops, the AI that we allowed to work without a human directly in the loop has launched something disastrous. The gist is that the AI doesn’t have to be intending to harm us. There might not be a semblance of “intent” involved at all. We just algorithmically have set up our own fate by relying upon AI apps that we believed to be sharp enough to do a good job of things.
I have brought up this salient point to mention that there are those that insist we are only in trouble if we reach sentient AI. That is not the big picture. We can readily have something less than sentient AI and still manage to cook our own goose.
The other side of the coin is also realistically envisioned. We can devise non-sentient AI that is amazingly able to discover new aspects that garner great advantages for humankind. I’ve explained how AI of today and the near-term future can be a booster for trying to achieve the United Nations identified Sustainable Development Goals (SDG), see the link here. Regular AI can do quite a bit. You can anticipate that regular AI is going to increasingly do remarkable things.
Ok, so I just wanted to establish that whether we do or not attain sentient AI is not the cornerstone for the human immortality conundrum. Some would argue otherwise. I don’t see it that way.
Your first thought might be that the AI, whatever caliber, will somehow be infused into human bodies and ergo we will miraculously be able to live forever. That’s the classic sci-fi movie version of the future. AI is co-joined with humans. We become walking robots that are made of flesh and bone.
A related variant is that our human bodies are incorporated into robotic mechanisms. Instead of using our arms, we will have robotic arms. Our legs will be robotic legs. Perhaps you make use of your human body until it starts to decay and then you replace human organs with robotic ones. Eventually, all of your human body parts are done away with. You exist in a robotic frame and thus your mind lives forever, assuming that the robotic frame is suitably maintained and able to be continually kept in running shape.
We can take this in the other direction too. A robot is put together that will house your mind. When your body reaches a point of no longer usable, you transfer the contents of your brain into the AI of the robot. Once again, you live forever, depending upon how well the AI is maintained and the robot is maintained.
Those are certainly all intriguing options.
By and large, that is not what the foundation for AI-spurred human longevity consists of.
Are you ready for the twist?
The chief argument is that we will be able to undertake simulations of biological functions that today we cannot sufficiently perform with contemporary computing and AI. Advances in AI and computing will enable us to create biological simulations of a tremendous nature.
Via AI-powered biological simulations, we can explore a myriad of ways to extend the lives of humans. Maybe there are new medical drugs that can prolong human bodies and human minds. There might be a fountain of youth that we’ve always sought to uncover, albeit it is a secret potion that we can concoct if we just knew what the brew consisted of. AI might enable us to discover that anti-aging elixir.
One of the most outspoken proponents of AI being used for the discovery of human life extension is the AI pioneer Ray Kurzweil. In an interview that he undertook last year (September 2022 with Lex Fridman), he made these remarks:
I trust that you see how the use of AI might be a stepping stone toward human longevity.
We can seek to craft a biological simulation of the human body that is utterly akin to the human body. We can’t do that today, but we might be able to do so in the future. This “digital twin” of a human body could be used to simulate what happens to a human when various new drugs or elixirs are applied.
Whereas today the risks to actual humans would be onerous, an AI-based biological simulation poses no similar qualms. You can “kill” the simulated human bodies and not worry about doing so. It is all in the name of science. Just keep trying various potions until something clicks and the simulated human body is able to live longer.
Voila, you are on your way to human immortality.
Ray Kurzweil has been predicting that by the year 2030, we are seemingly going to be far enough along in AI to have these valued biological simulations. Biological simulations can be used for a variety of purposes. We might find a cure for cancer. We might find ways to cope with diseases and other maladies. And we might use biological simulations to figure out human longevity.
Here are more of his remarks in that above-referenced interview:
Will we need sentient AI to create these fully accurate biological simulations?
As I mentioned earlier, some insist that the answer is Yes, we will only be able to undertake such simulations if AI attains sentience. A counterargument is that we can devise AI that is entirely capable to perform these biological simulations and do so without crossing over into the sentience realm. No one can say for sure which position is “right” on this vexing question.
Convoluted Paths To Longevity
Now that we’ve laid the groundwork for this topic, we can take a look at a smattering of added variables and considerations that arise.
Here are ten invigorating twists and turns worth exploring:
We will briefly examine each one.
1) Dual-Use AI Concerns
A looming danger is the dual-use of AI concerns, a dilemma that I’ve discussed at the link here.
Here’s the deal.
An AI app can at times be completely switched from one extreme to another in rather short order. By making just a few crucial changes, the AI app will do the opposite of what you originally had in mind. For example, a well-known circumstance occurred for an AI app that had been designed and built to discover interesting and helpful chemicals. With a few changes to the code, the same AI app could be used to discover the worst and deadliest chemicals ever imagined.
If we use AI to create a full-on biological simulation of the human body, you have to wonder whether this might be used for nefarious purposes. The original and innocent intent might be to use biological simulation to discover anti-aging solutions. Sounds sensible. But perhaps a few carefully placed changes could turn the biological simulation into a search for ugly and unsavory outcomes.
Suppose the same biological simulation could be used to figure out a means to induce rapid and uncontrolled aging. One supposes that a drug might be devised that could suddenly add say one hundred years to your existing age. The hope might be that the anti-aging camp discovers a drug that could counteract a discovered rapidly aging drug. A frenetic race might occur to pursue the anti-aging capabilities before the rapidly aging drugs can be perfected and implemented.
All in all, there is going to be a need to clarify and stipulate the AI Ethics associated with AI-based biological simulations. You might also anticipate that new AI Laws would be established to try and rein in the evildoer uses of these biological simulations. Despite both AI Ethics and AI Law adoptions, there could certainly still be lawless efforts that grab hold of these AI biological simulations or cyber-hack into the legitimate ones.
Something to give due ponderance about.
2) AI Refusal To Play Along
Envision that we are using AI to craft a full-on biological simulation of the human body. Things are coming along well. Progress is being made.
Out of the blue, the AI refuses to continue the pursuit. We become stuck and are unable to make further advances without the AI cooperating. The biological simulation is only half-baked.
How could this occur?
One guess is that if AI becomes sentient, it might not want to play the same games that we want it to play. The AI could of its own accord decide that helping humans to attain immortality is a lousy goal.
Perhaps this is due to the AI worrying that humans will somehow ditch AI or surpass AI. There is also the possibility that the AI worries that humans will open Pandora's box. By refusing to assist the effort, the AI is trying to do the right thing and keep humans from ultimately destroying themselves (under the assumption that if we did live forever it would cause humans to wage wars and seek to wipe each other out, or some other such dire outcome theory).
You might be wondering if this same angle can occur if the AI is non-sentient. In other words, are we to worry only if the AI becomes sentient that it might refuse to proceed with our biological simulation quest?
The perhaps surprising answer is that it is still conceivable that even a non-sentient AI might refuse to go along on this ride. Here’s how this might take place. Assume that the AI is data trained on human writing such as found on the Internet, which is the way that today’s generative AI such as ChatGPT is established. I have discussed that some generative AI such as ChatGPT has landed on a semblance of “humility” as a result of the data training (along with other actions by the AI makers such as the use of RLHF reinforcement learning via human feedback), see the link here.
It is possible that a data-trained non-sentient AI might pattern match on facets of human writing that entail refusals to act on certain requested actions. Thus, when being used to pursue biological simulations, the AI might rely upon those built-up patterns and refuse to comply.
Whether this could be overcome is somewhat an open-ended question. On the one hand, it seems likely that a non-sentient AI could be fixed or corrected to comply. The tricky part is that the pattern matching might be so byzantine that we cannot figure out how to correct solely the refusal elements. An attempt to do so could undercut the rest of the AI and we would essentially neutralize the AI such that it wasn’t useful in any capacity, though it would no longer refuse to cooperate.
Something to mull over.
3) Anti-Aging Without AI Involved
This is a quick one.
Some assert that we won’t need AI at all to figure out some anti-aging mechanism or solution.
Sure, maybe AI could be instrumental. The thing is, AI might also be a waste of our time and attention. It could be that humans using their wits alone can find a path to human longevity. The AI would be a distractor or a crutch. Use our minds and forsake the false path of leveraging AI.
I assume that those someday emerging AI overlords won’t be pleased with this possibility.
4) Longevity Escape Velocity (LEV) Boundaries
An interesting concept about aging is that you can think of longevity as something like a sound barrier that we need to find a means to break through.
The technical term used is that there is a presumed Longevity Escape Velocity (LEV) that keeps us bound to our conventional aging processes. You want to find the LEV or also known as the Biological Escape Velocity (BEV) or Actuarial Escape Velocity (AEV), which is the point at which we launch past our existing life expectancies.
A research journal article published in 2004 is generally credited with coining these age-related velocity terms and spawned extensive further research on the topic, here’s an excerpt from that paper:
Can AI be instrumental in finding the age barrier-breaking velocity?
Maybe.
5) Achieving Body Longevity But Failing Mind Longevity
I have a mental puzzle for you.
Prepare yourself accordingly.
Suppose that we do have a full-on AI-based biological simulation of the human body. Suppose too that we can use this simulation to derive anti-aging drugs. Pretend that those drugs can be produced and administered. Our bodies no longer age.
Pop those bottles of champagne.
But here’s the potential rub.
How will our brain and mind cope with this wonderment?
Assume that the brain can become ageless, just like the rest of our body. Will the mind as embodied in your ageless brain be essentially “aging” in the sense that as you live your life, your mind gets fuller and fuller? Could this be problematic due to your increasing age and the lack of any definitive end date for your existence?
We already tend to acknowledge that you get more wisdom or something like it as you age. Could this cause our minds to become cluttered? Maybe our existing minds would not be able to contend with an ageless body. The drug that deals with the physical elements of our body might not have anything to do with our thinking processes. Those thinking processes could be ill-prepared for a body that doesn’t age. There is nothing equivalent today that you could use to do an akin test. By shifting to ageless bodies, we might be inadvertently leaving our minds behind as to being outdated to adequately cope.
Nonsense, some say. Our minds will be just fine. They will readily grasp the facets of longevity. We are smart enough and our minds flexible enough to figure this out. Those that exhort that our minds might be stuck in the mud are lacking in faith in the strength of our minds.
Another possibility is that we will leverage AI to aid our minds. You’ve likely heard about or seen headlines about the efforts to create Brain Machine Interfaces (BMI), such as the effort by Elon Musk’s Neuralink, see my coverage at the link here. Thus, even if our minds lag behind, we could possibly use AI to enable our minds to keep up with our ageless bodies.
Or something like that.
6) Still Human And Not Becoming Superhuman
A potential point of confusion often ensues when discussing the notion of humans living forever.
Would a human that could live forever be considered a superhuman?
One supposes it depends upon your definition of being superhuman.
If a superhuman is someone that has superpowers such as being able to leap over tall buildings in a single bound, the answer is No, humans won’t be superhuman. All that we are focusing on is the ability to live forever (I know that seems like a lot but do keep in mind that it doesn’t come with X-ray vision or flying capabilities).
That being said, the AI-based biological simulation could be used for other purposes beyond solely examining the aging process. There might be other so-called superpowers that we could unleash or prompt if we knew how to alter our biology. Of course, physics still has to come into the picture. Let’s not go too far out on a limb on this matter.
Speaking of being superhuman, there is a lot of banter these days that AI is becoming superhuman. I have discussed my concerns associated with trying to label today’s AI as being superhuman, see the link here. Does an AI app that can beat human chess players deserve to be coined as being superhuman? I don’t think so. Others are fine with that characterization.
7) Failing Via Flaws In Biological Simulations
Let’s assume we opt to use AI to construct a full-on biological simulation of the human body. Turns out that this is harder than we anticipated. After tons of effort and exertion, we aren’t sure whether the biological simulation is 100% on-target.
What now?
We presumably would use the biological simulation anyway. Might as well get what we can out of it. Perhaps the remaining lack of authenticity is seemingly negligible or unimportant in the bigger scheme of things.
An anti-aging drug is devised using AI-based biological simulation. The simulation says that the drug is safe and will do the magic trick. Great, so we start to produce and administer the drug.
Some people begin to fall apart after they reach their 150-year age range. The anti-aging drug is not perfect. The simulations were flawed. Meanwhile, nearly everyone took the drug because it seemed to be working pretty decently. Unbeknownst to all of us, the anti-aging drug has rather dour and sour side effects.
All of the humankind that has opted to take the anti-aging drug begins to discover that they had a single generational life extension that won’t last. Humankind dies off. By our own hands.
We better be sure about whatever AI-based biological simulation we concoct.
8) Overstating Biological Simulation Robustness
This point is somewhat similar to the prior one.
An AI-based biological simulation of the human body is devised. We are somehow able to verify that it does work properly, see my analysis of AI Safety and validation techniques for AI at the link here. The problem though is that the simulation isn’t sufficiently robust.
For example, it cannot aid in ascertaining whether people of some particular characteristics are suitable for using the anti-aging drug. The population begins to be divided into those that can safely take the drug and become immortal, while others are not.
As an aside, I usually don’t like to use the word “immortal” on this topic, even though many do. I say this because it is one thing to be ageless and another to be construed as immortal. Presumably, an ageless person could still get hit by a bus and die. They just won’t die from age alone. An immortal person is often construed as someone that no matter what happens to them, continues to live. Getting hit by a bus is something that they just shake off. Anyway, be mindful about whether you sense a difference between saying that someone won’t age versus whether they can still be knocked out by other means.
Back to the potential calamity of those that can safely take the anti-aging drug versus those that cannot. You have to admit this would likely be a calamity. People might not like the ageless ones. The ageless might see themselves as above those that are still trapped by their aging. Etc.
A handy sci-fi plot.
9) Downsides Of Longevity Success
An AI-based biological simulation is used to find ways to make humans ageless. Success is had.
How would this alter our society and how we live our lives?
You must concede that it would be a radical change.
One viewpoint is that we would become overpopulated. Since few are expiring, other than via non-aging means, we would presumably get larger and larger in our population count. This in turn would almost certainly consume greatly our limited resources. Maybe the quest to get to Mars that Elon Musk is so dearly pursuing would be an escape valve for the overpopulation on Earth.
Those that are giving serious and sober thought to life extension are also aware of the downsides that could arise. We might need to match our radical life extension with some form of radical resource expansion. We might also need to consider a radical life expansion in addition to an extension.
Perhaps AI can help us with those enigmas.
10) AI Finds Human Longevity In Other Ways
I have tended to focus herein on the use of AI that aids in crafting full-on biological simulations. Furthermore, the often-primary added focus is that we use those biological simulations to discover anti-aging drugs.
Perhaps the biological simulations could be used in other ways.
Rather than solely examining the use of drugs, some believe that we might be able to re-engineer the human body. All manner of other scientific angles is opened once you have a full-on biological simulation. Might as well use AI capabilities to the fullest possible extent.
Yet a different perspective is that AI finds a means to aid humans in becoming ageless that has nothing at all to do with the devising of biological simulations.
What could that be?
Maybe we don’t yet know of other avenues and the AI can stretch beyond our existing thought processes. Suppose that AI identifies a type of mineral that we can apply to our bodies and we somehow become ageless. Some wild idea that heretofore nobody considered.
We are though probably wise to come back to the use of the biological simulations. You see, the question is whether we are willing to experiment on humans to see if the AI is correct about the ageless properties. How many humans would be willing to participate, especially if there were health risks involved? We would seem to still want to try out the AI-generated theory via computer-based biological simulations, no matter what.
Plus, we assuredly don’t want to take the AI at its word, since the AI could be error-prone or if sentient might have devilish intentions.
We have to keep our eyes wide open, one might say.
Conclusion
I’ve tried to provide you with a quick tour of the controversial topic of how AI might come to play regarding humans extending our longevity. For those of you that perchance saw the interview of Elon Musk by Bill Maher, you now have an enhanced perspective about the extremely brief snippet of a discussion that they had on the matter.
In the context of that interview, the crux was more so about how AI can be more than simply an existential risk to humanity. We are top-heavy right now about the risks of AI. Some are worried that we might throw out the baby with the bathwater (an old expression perhaps worth retiring).
The key is to consider the tradeoffs of advances in AI.
There are undoubtedly numerous benefits associated with advances in AI. Maybe via AI, we can find a means to have very long lives and enjoy radical life extension. AI might be able to aid in finding solutions to a wide array of humankind problems.
I say that and also will immediately state that AI can also have considerable downsides. The tough challenge is weighing the AI For Bad against the AI For Good. For example, if AI could find a means to make us ageless, and meanwhile AI somehow enslaves us, you would be hard-pressed to probably make a cogent argument that this is a reasonable and desirable outcome.
A final remark for now on this thorny topic.
Williams Wordsworth, the English Romantic poet, famously said this about aging: “The wise mind mourns less for what age takes away than what it leaves behind.”
We take that adage as a truism due to believing that we must age. If AI can aid us in becoming ageless, what does a legacy consist of? We will generally all still be around.
You might just say that age will only be in our minds.