Understanding WHY Anthropic CEO Dario Amodei is claiming that Cancer, Alzheimer’s, Heart Disease, Mental Illness will be Relics of the Past in Less Than A Decade

People are interpreting Dario Amodei’s comment that AI will cure Cancer, Alzheimer’s, Heart Disease, Mental Illness in less than a decade in 4 main ways

  1. He’s a con man – Trying to pump Anthropic IPO
  2. He’s a dumbass who without any actual experience in a medical research company is making assumptions about medical research
    1. Hubris of AI doing so well in replacing software development engineers has made him (and other AI researchers) think that they can replicate that in all fields
    2. Very similar to how President Trump taking Venezuela easily convinced US and Trump that Iran would fold easily too
  3. Dario Amodei has gone crazy and has started believing absolutely crazy things out of madness
  4. It is actually going to happen
    1. That Anthropic has some inside knowledge about Medicine and using AI in Medicine that shows it that it’s going to cure everything in 10 years
    2. We should consider this perspective too. Regardless of how crazy it seems

We would argue that none of these are the real reason Dario Amodei is making a fool of himself.

When someone says something completely crazy, it is sometimes helpful to understand what their real motivations are.

Perhaps Dario Amodei’s REAL MOTIVATIONS behind claiming AI will cure Cancer, Alzheimer’s, Heart Disease, Mental Illness within 9 to 10 years (Less Than A Decade) are Very Interesting Motivations

Here are the REAL motivations (our assumptions) for Dario Amodei going off his rocker

  1. If Anthropic doesn’t raise $100 billion in 2026 for its 2027 expenses
    1. Anthropic might go bankrupt
    2. If Anthropic doesn’t raise another $100 billion in 2027 for its 2028 expenses
    3. Anthropic will probably go bankrupt
    4. A non frontier lab with huge profits, such as Microsoft or Apple or Google or Facebook, will buy Anthropic
    5. Instead of becoming ‘Biggest company in the World’ and ‘Most Important Company in the World’
      1. Anthropic will become a subsidiary of a Hyperscaler or of Apple
    6. Instead of Dario becoming AI’s Chosen God King and ruling over the world
      1. Dario would just be another Founder who burned bright and then ended up doing Rest and Vest at a large Silicon Valley Tech Firm
    7. Instead of the Twin Religions of Effective Altruism and AI Is Our New God becoming the dominant religions/belief systems in the world
      1. They would go back to purgatory as inconsequential belief systems with twin grand failures of Sam Bankman Fried and The Amodei Siblings
      2. Effective Altruism barely recovered from Sam Bankman Fried effectively altruizing $8 billion from 1 million FTX investors
      3. Surely, if Anthropic and its EA Super Heroes mess up, there will be no way back for Effective Altruists
  2. Anthropic and OpenAI’s earlier strategy of – Give us all your money or AI will kill everyone – was not working
    1. OpenAI reduced its fund raising from $30 billion at $1.5 trillion valuation to $11 billion at $1.25 trillion valuation
    2. Anthropic has had to reduce its valuation from $2 Trillion to $1.4 Trillion
    3. This clearly shows that the earlier strategy of Doom Mongering is not working
    4. If you cannot convince people that AI will kill them, then the only other equally strong message is …
    5. AI will make you immortal. If you give us all your money
      1. AI will cure cancer so you can live longer
        1. Eventually, you can live forever as Cancer and Ageing are the only two remaining Great Enemies of Immortality for Humans
      2. AI will cure all mental diseases
        1. Not only can you live forever, you can live happily ever after
    6. This is an overwhelmingly positive message – Just invest in Anthropic’s IPO and you will live FOREVER and you will live HAPPILY EVER AFTER
      1. You might not have a job as AI becomes better
      2. You will, however, have immortality
  3. Anthropic and Dario Amodei are painfully aware that End 2026 and Early 2027 might be the ONLY windows to raise crazy amounts of money at a crazy valuation
    1. The stock market might go down. Killing Anthropic’s window to raise a crazy amount of money. Consider all the things that might spook the market
      1. The most powerful CEO in AI might start shagging his business partners and his employees
        1. That might crash Nvidia stock
        2. That might in turn bring down the elegant and beautiful circular financing mega-structure Mr. Shag-A-Lot has created
      2. Coreweave and/or Oracle might fail their debt obligations
        1. That too would bring down the Circular Financing Mega Structure
      3. Data Centers and Power Grids might be unable to expand fast enough to support the crazy amounts of growth Anthropic and OpenAI need to experience to keep the AI Party going
        1. Anthropic has $800 to $900 billion of commitments in the next 10 years
        2. OpenAI has between $500 to $800 billion of commitments in the next 10 years
        3. If either company fails, what happens to all the companies whose existence and/or growth depends on Anthropic and OpenAI?
          1. This is especially important for NeoClouds who have no revenue source other than AI
          2. This might even be an existential threat for Oracle, which has gone all-in on AI Data Centers
      4. War in Iran might re-start
      5. War between Ukraine and Russia might spread to all of Europe
      6. China might decide it really wants a Huang Huang Bang Bang with Taiwan
    2. Anthropic might never again be in such a strong position
      1. The mixture of token-maxxing, OpenAI not focusing on Enterprise until 2026, Anthropic’s newest model being quite far ahead of OpenAI’s newest model is quite delicious for the Anthropic family
        1. We might never again see a situation like this
        2. Already we have seen that OpenAI earnings accelerated in Q3 while Anthropic’s earnings did not grow as fast in Q3 as they had in Q2
      2. Now is the PERFECT and perhaps ONLY time for Anthropic to IPO as undisputed frontrunner among Frontier AI Labs
    3. Open Source might make big advances and close the gap
      1. Some people think Open Source is just 6 months behind
      2. Other people think Open Source is just 12 months behind
      3. A few people think Open Source can never catch up to OpenAI and Anthropic
      4. What none of these people can argue against is the REALITY that Open Source has taken a large part of the market
        1. Companies such as Revolut have shifted most of their spend to in-house Open Source based AI models
        2. OpenRouter’s measuring and metrics clearly show that Open Source is slowly and surely taking over most of the AI market

Anthropic is in a race against Extinction (its own).

Dario Amodei is behaving with the urgency of a CEO who knows that if he can’t conjure up $900 billion in the next 7 to 9 years, his company will run out of time and go bankrupt and get acquired by one of his mortal enemies.

Oh, the Horror of handing over the Care and Nurture of the AI God to a non believer.

If he says out loud that Anthropic needs massive funding to survive – It will kill confidence and that will increase the chances of failure.

So he must pretend to be in a position of massive strength – Pretend that complete domination by Anthropic is inevitable.

Sun Tzu in ‘The Art of War’ -> Appear Strong When You Are Weak

Anthropic and OpenAI tried the stick approach i.e. Funding them is necessary for security and for safe development of AI.

It did not work.

Now Anthropic is trying the carrot i.e. Live Forever, Be Free of Cancer, Be Free of Mental Health Issues.

Dario Amodei must raise money for Anthropic and if Fear is not working, then perhaps Greed will.

Perception is Reality, unless you believe in Generally Accepted Accounting Principles

If Anthropic were really strong financially, it would just use Generally Accepted Accounting Principles. It would use Annual Recurring Revenue instead of Annual Run Rate.

If Anthropic were immensely strong as a company, it would talk about what developments are going to happen in 1 or 2 years, instead of promising the Singularity in 2027 and A Cure for Cancer within 10 years.

We are seeing a CEO who is fighting for the very existence of his company. The threat to the existence of humanity and the promise of immortality for humanity are the best things to use as the stick and the carrot.

Will AI actually cure Cancer, Alzheimer’s, Heart Disease, Mental Illness within 10 years?

The chances are much closer to Zero Percent than to 100% Percent.

Is it possible that it will happen within 10 years? Anything is possible.

Is it likely? Absolutely Not.

At some level it is just sad that this dude Amodei has fallen to the level of completely shameless storytellers like the ‘Data Centers in Space’ storytellers.

Archbishop Amodei of the Church of AI, you have a great product. Let it survive on its own merit. By being so desperate and making all these absurd claims you are not helping your IPO.

The Reasons Curing Cancer in 10 Years is a Pipe Dream are Rather Simple

  1. You can’t just understand a field very quickly
    1. As opposed to Coding which was all already built on computers and done on computers, medicine involves the human body
    2. As opposed to Coding where the programmers were extincting themselves by allowing OpenAI and Anthropic to use all their code and study what they did, there is no easy way to get all the steps of the process for healthcare
      1. Have you seen doctors and patients covered in cameras and monitors, uploading all that information for an AI to process?
        1. Almost no one has because almost no one is doing this
        2. Contrast that with programming where immense number of people are sharing their code bases with AI companies and using AI coding tools
          1. For computer programming we have IMMENSE amounts of data and Almost Immense amounts of fresh, new data
          2. The AI can keep improving as it never runs out of more code to learn from
      2. That data gathering process for medical areas has not even started yet
        1. Just the data gathering might take 5 to 20 years
  2. After understanding a field and gathering data – You can’t just find great solutions very quickly. You need world class researchers to direct all the research. Without enough data and context, even Agents possessing Super Intelligence can’t do anything. You need actual medical researchers and input from actual healthcare professionals. You need world class professionals
    1. In parallel with the 5 to 20 years to gather data, we would need another 5 to 10 years to process and understand that data and build solutions. A lot of the work might not be possible in parallel
      1. We need specialized doctors to guide everything
      2. We need specialized programmers to build the AI systems
      3. The overlap is almost zero
        1. Imagine the typical specialized surgeon talking to the typical coding savant
        2. They have to understand each other’s language and mindset, not just the problem they are solving
          1. How many programmers do you know with great bedside manner?
        3. With coding everything was THE SAME PEOPLE
          1. Now we have two sets of domain experts who don’t understand each other’s languages or domains
  3. You can’t just ‘invent’ a cure and start giving it to people
    1. After all this data gathering and data processing and research we start arriving at possible solutions
    2. Can we just start giving our ‘might cure cancer’ solution to human beings?
    3. No, we cannot
  4. Testing on Animals takes years
    1. AI Companies want you to believe that you can have 1 million agents running in parallel and exploring every possible cure
    2. How do you test it given you don’t have 1 million rats or 1 million monkeys to test it on?
    3. Can you make the diseases and the cures progress ‘faster’?
      1. Cancer is not exactly a fast moving disease
    4. It would be a miracle if anyone can get the animal testing phase done within 1 or 2 years
    5. In reality it might takes 2 to 10 years to test on animals
      1. Even if we do this in parallel with earlier stages, it will still add 1 to 5 years
      2. Not to mention the ethics and issues with testing AI created experimental solutions on animals
  5. There are three stages of testing for humans
    1. We’ve tested the cures on animals. Now we progress to Human Testing
    2. There are three phases
    3. You cannot magically create millions of humans with varying strains of cancer and start testing thousands of cures in parallel
    4. Each phase takes years
    5. Human Testing alone can take 5 to 10 years.
      1. It cannot be done in parallel with animal testing.
      2. It cannot be done in parallel with data gathering.
      3. It cannot be done in parallel with finding cures
      4. When Anthropic CEO Dario Amodei says – We can cure cancer in less than a decade
      5. Either he thinks Cancer Clinical Trials will be done on AI agents
      6. Or he is completely delusional and thinks all the Governments will drop human testing requirements
    6. After the initial period to find a cure and test it on animals, there are human clinical trials
      1. 5 to 20 years data gathering
      2. (Some in Parallel) 5 to 20 years processing data and research
      3. (Some in Parallel) 1 to 10 years of testing on animals
      4. (Unlikely to be in Parallel) 5 to 10 years of clinical trials on humans
      5. That means it would take 12 to 80 years for the cure to be confirmed
  6. Even after testing you need years to get it to people
    1. Let’s say we get everything done within 12 to 80 years. Perhaps 15 years to be extremely optimistic
    2. How long is it going to take to get it to a significant number of people
    3. To confirm it is safe in actual usage?
    4. It might be possible in 5 to 10 years, if everything goes perfectly
    5. In reality, it might take 20 to 50 years
    6. On top of the 12 to 80 years to find and test a cure and arrive at a cure
    7. We have another 20 to 50 years for the cure to actually make Cancer a ‘relic of the past’ by getting it to all people who need it
    8. Our Reality Based Time Estimate says that it would take 32 to 130 years to actually make Cancer a ‘Relic of the past’
      1. Assuming Everything Goes Reasonably well
  7. How on Earth are you going to get it to poor people?
    1. Something that should not be overlooked is that these Effective Altruists are not going to give their products to poor people for free
      1. From their alleged very high operating margins it is clear that their main customers are the very profitable and immensely rich trillion dollar hyperscalers
      2. We do not see any ‘For Free’ donations to Small Companies or to Solopreneurs
        1. Everyone has to pay
    2. Effective Altruism fails if they can’t first take money from people
      1. They are only taking money from people to (at some unknown point of time in the future) give it away effecitvely
      2. However, they are not giving away ANYTHING to anyone at the moment
    3. So the poor people with Cancer would not be able to afford it, and would have to first wait for Anthropic to become so rich that its ‘At some point of time we will give away the cancer cure to poor people for free’ clause kicks in
      1. Let us hope the new Effective Altruists are not like the old Effective Altruists (FTX) who were only doing Effective Altruism from FTX accounts to their own personal accounts
      2. Because that would mean that poor people might NEVER get the cure and that Cancer might never ever be a relic of the past

Finally, it is worth noting that the cost of failure is VERY HIGH in Medicine and Healthcare.

If an AI hallucinates and makes an error in an App, you don’t lose an arm. If an AI makes a mistake and breaks your company website, it doesn’t cause a death.

People are being glib when they say – That person was going to die anyways from Cancer. Doing AI trials is free of risk.

Different cancers have different rates of survival. Most cancers now can be treated and/or managed. Very few human subjects would be willing to take the risk that a computer error means instead of a miracle cure they get a faster death.

In Summary, even with Extreme Optimism, A Cure for Cancer is 32 to 130 years away

I would be delighted to be wrong. In our hopefulness that cancer will be cured by AI, we should not drown in the delusion that A Cure for Cancer is 10 years away.

Even the most optimistic estimates, which assume AI can very quickly understand and research cancer and find a cure, gives us an estimate of

  1. 12 to 80 years to get an actual cure for cancer and test it
  2. 20 to 50 years to get it to everyone and actually make Cancer a ‘relic of the past’

That’s 32 to 130 years.

Anyone with even an iota of honesty would not say – In Less than a Decade Cancer would be a Relic of the Past.

Even if you think you are building a new AI God to rule over the Human Race, you still should not claim your AI God will make cancer a relic of the past in less than a decade. None of the existing Gods have been benevolent enough to bequeath humanity with a cure for cancer. There is no evidence or indication that the AI God will be so magnanimous.


Leave a Reply