Yesterday in Stockholm, Geoffrey Hinton and John Hopfield received the Nobel Prize in Physics for foundational work on artificial neural networks, and Demis Hassabis and John Jumper shared half of the Chemistry prize for AlphaFold's protein structure predictions. It is the first time AI research has been honoured this way, and it caps a year in which the technology changed character.

In September, OpenAI released o1-preview, a model trained to reason through problems step by step before answering, with large gains on competition mathematics and science questions. Coding assistants moved from autocomplete to writing and editing whole features. In October, Anthropic's CEO Dario Amodei published "Machines of Loving Grace", a long essay imagining AI compressing decades of scientific progress into a few years. In March, the philosopher Nick Bostrom, best known for warning about AI risk in Superintelligence (2014), published Deep Utopia: Life and Meaning in a Solved World, a book about the opposite problem: what if things go well?

Whether these predictions prove right or wrong, they have made an old philosophical question feel practical. If machines can do much of what we do, and do it better, what is the point of us doing it? Where does purpose come from when you are no longer needed?

This post is a year-end guide to that question. It is not a forecast. It is a look at what philosophy and psychology already know about meaning, and what that suggests for a world with more capable machines in it.

1. Three words that get mixed up

Concept Meaning Question it answers
Happiness Feeling good and judging your life to be going well "How is my life going for me?"
Purpose Having goals that direct your activity over time "What am I trying to do?"
Meaning Psychologists Frank Martela and Michael Steger (2016) identify three facets: coherence (life makes sense), purpose (direction) and significance (life is worth living and matters) "Does my life make sense, and does it matter?"

These come apart. A life can be pleasant and feel meaningless, or hard and deeply meaningful. Roy Baumeister and colleagues found in a 2013 study that happiness was associated more with getting what you want and need, while meaningfulness was associated more with giving, with linking past and future, and with expressing yourself, even at some cost to happiness.

The AI question is mostly about purpose and significance. If your purpose has been "be good at this work", and significance came from "people need what I do", then a machine that does the work threatens both.

2. Keynes saw this coming in 1930

In the depths of the Great Depression, John Maynard Keynes wrote a short essay, "Economic Possibilities for our Grandchildren". He predicted that within a hundred years, technical progress and capital accumulation would make people in rich countries four to eight times better off, solving "the economic problem". People might then work perhaps fifteen hours a week, and humanity would face its "real, permanent problem": how to use its freedom from pressing economic cares "to live wisely and agreeably and well".

He was right about the growth, roughly, and wrong about the hours. Working time in rich countries fell through much of the 20th century and then largely stopped falling. Many professionals work more than their grandparents did. Economists offer several reasons: rising wants, status competition, the rewards of high-paying work, and the fact that for many people work became a source of identity rather than only a means of earning a living.

Keynes also worried about the transition. He expected "a general nervous breakdown" among those who had no talent for leisure, comparing them to the wives of the wealthy classes of his day, deprived of useful tasks. He did not think freedom from necessity would be easy.

The 2024 debate is Keynes's question with a faster clock.

3. What work gives us besides money

In 1982, social psychologist Marie Jahoda, drawing on her studies of unemployment going back to the 1930s, argued that employment provides latent functions beyond income:

  1. Time structure: days and weeks have a shape.
  2. Social contact outside the family.
  3. Collective purpose: participation in goals larger than oneself.
  4. Status and identity.
  5. Regular activity.

Research on unemployment supports her. The loss of wellbeing from becoming unemployed is consistently larger than the loss of income alone would explain, and it often persists after people find new jobs.

This is the strongest reason to take the meaning problem seriously. Even if a world with capable AI solved income, through whatever economic arrangement, it would still need to supply structure, contact, shared purpose, identity and activity. Some of these are easy to find outside paid work. Others, especially collective purpose and identity, are harder.

4. What the philosophers offer

Philosophy has thought about meaning without economic necessity for a long time, often because philosophers wrote for, or were, people of leisure.

Aristotle: excellence in activity

For Aristotle, the good life (eudaimonia) is "activity of the soul in accordance with virtue", exercised over a complete life, among friends and in a community. Two features are relevant here.

  • Meaning lies in exercising capacities well, not in being the best at them or in being needed. A good flute player flourishes in playing well, whether or not someone else plays better.
  • The highest activities are valued for their own sake. Aristotle ranked contemplation, the free exercise of understanding, above practical work precisely because it serves no further purpose.

On this view, a machine that does a task better does not remove the value of you doing it well. It removes only one reason (usefulness) among several.

Camus: meaning in the struggle

In The Myth of Sisyphus (1942), Albert Camus took the most pointless task imaginable, rolling a rock up a hill forever only for it to roll down, and concluded: "The struggle itself toward the heights is enough to fill a man's heart. One must imagine Sisyphus happy." Meaning, for Camus, is not given by the cosmos or by the outcome. It is in how we meet our situation.

Frankl: three sources of meaning

Viktor Frankl, a psychiatrist who survived Auschwitz, wrote Man's Search for Meaning (1946). He identified three sources of meaning:

  1. Creating a work or doing a deed.
  2. Experiencing something or encountering someone, above all, love.
  3. The attitude we take toward unavoidable suffering.

Only the first is threatened by capable machines, and only partly: a work created by you, for someone, still carries meaning a better machine-made version does not. The second and third are untouched.

Susan Wolf: subjective and objective

The contemporary philosopher Susan Wolf, in Meaning in Life and Why It Matters (2010), offers a useful formula: "meaning arises when subjective attraction meets objective attractiveness". You need to care about the activity (subjective), and it needs to be worth caring about (objective): relationships, creative and intellectual work, helping others. Neither alone is enough: loving something trivial is not meaningful, and doing something valuable that you are alienated from is not either.

The AI question for Wolf would be: does something remain objectively worthwhile when a machine can do it? For relationships, care and personal growth, clearly yes. For some kinds of work, the answer may change.

Bernard Suits: utopia is a game

The most surprising resource is a playful 1978 book by Canadian philosopher Bernard Suits, The Grasshopper: Games, Life and Utopia. Suits defines playing a game as "the voluntary attempt to overcome unnecessary obstacles". He then argues that in a utopia where all instrumental needs are met, the only remaining activity worth doing would be game playing in this broad sense: freely choosing challenges for their own sake, from sport to science pursued for the joy of discovery.

Suits's point is that we already know how to find meaning in activity that is not necessary. Every marathon, chess game, amateur choir and garden is proof.

5. The 2024 contributions: Bostrom and Amodei

Bostrom's Deep Utopia

Bostrom's Deep Utopia (March 2024), written in an unusual form mixing lectures, dialogues and fiction, takes Keynes and Suits to the limit. He imagines a "solved world", where technology can not only do all economically useful work but even achieve, more efficiently, things we now do for their own sake, such as getting fit or learning. If a pill can give you the fitness or knowledge, what is the point of the training or the study?

Bostrom's analysis, simplified:

  • Many sources of value survive. Pleasure, rich experiences, relationships, aesthetic appreciation and play do not depend on being useful.
  • Purpose is the hardest to preserve. When effort is never needed to achieve outcomes, purposes based on necessity disappear. Some can be recreated: through games, through commitments and through goals that specifically require our own efforts (being there for a friend cannot be outsourced).
  • The question is what we value, not whether we can find activity. A solved world forces us to decide which of our values were really about outcomes and which were about the activity itself.

Bostrom is cautiously optimistic, though he leaves many questions open. The book's value is less in its answers than in showing that the meaning problem is a real philosophical problem, not only an economic one.

Amodei's Machines of Loving Grace

Amodei's October essay is mostly about what AI might do for biology, neuroscience, economic development and governance. Its last section, "Work and meaning", addresses our question directly. He argues that meaning and economic usefulness are separable, writing: "I think meaning comes mostly from human relationships and connection, not from economic labor." He notes that people already find meaning in activities a machine could do better, and that being the best in the world at something is not necessary for it to be meaningful.

He is more cautious on the economics, acknowledging that in the long run AI may become so effective and cheap that current economic arrangements no longer apply, and that society will need to work out new ones deliberately.

Both writers, from different starting points, converge on Frankl and Aristotle: relationships and the exercise of capacities for their own sake are the most durable sources of meaning.

6. Evidence from the board: chess and Go

We have a natural experiment in what happens to human meaning when machines become superhuman at an activity people care about.

Chess. IBM's Deep Blue beat world champion Garry Kasparov in 1997. Chess did not die. It is more popular today than ever, with tens of millions of active online players, boosted further during the pandemic. Human players study with engines, and top players are arguably stronger than ever. People still watch humans play, because what they care about is the human drama: preparation, nerves, mistakes and courage.

Go. In 2016, DeepMind's AlphaGo beat Lee Sedol 4-1. The shock was greater, because Go had been considered beyond machines for decades. In 2019, Lee retired, saying that even if he became number one, there was "an entity that cannot be defeated". That is the meaning problem in one sentence: excellence felt hollow once it could not be the best.

But that was not the only response. A 2023 study by Minkyu Shin, Jin Kim, Bas van Opheusden and Thomas Griffiths, analysing 5.8 million professional Go moves over 71 years, found that after superhuman AI appeared, human players began making significantly better decisions, and more novel ones. The machines did not end human play; they expanded what human players thought was possible.

The lesson is that meaning survives machine superiority when the activity was valued for itself, and is threatened when it was valued mainly for being the best. Lee Sedol's response and the wider Go community's response were both human, and both instructive.

7. For people whose identity is their work

For many readers of this blog, the question is personal. Software engineers, analysts, writers and designers built identities on skills that AI tools now perform, at least partially. It is reasonable to feel unsettled.

Some ways to think about it:

  • Separate the outcome from the craft. If you write code because you enjoy the craft, AI does not take that away, any more than calculators took away mathematicians' pleasure in proofs. If you write code for the outcome, AI may make you more effective at the outcome. Know which applies to which parts of your work.
  • Move toward the parts that require you. Judgment about what to build, responsibility for consequences, understanding people's real needs, mentoring and trust are much harder to delegate than producing artefacts.
  • Diversify your sources of meaning. Jahoda's list is a good audit. If work is your only source of structure, contact, purpose, identity and activity, any change at work becomes an existential threat. Building some of those outside work is wise regardless of AI.
  • Do not wait for the future to decide. The deep utopia scenarios are speculative and, if they arrive, distant. The tools of 2024 are powerful, but they still need people to direct, check and take responsibility for them. The practical question for next year is how to use them to do more meaningful work, not how to live without work.

8. A year-end reflection

December is a natural time to look back. Here are five questions, drawn from the thinkers above, to ask about your own year:

  1. Frankl: What did I create, whom did I love, and how did I bear what I could not change?
  2. Aristotle: Which activities did I do well and for their own sake, not only because I had to?
  3. Wolf: Where did what I cared about meet something genuinely worth caring about? Where did they come apart?
  4. Jahoda: If my job disappeared tomorrow, where would my structure, people, shared purpose, identity and activity come from?
  5. Suits: What unnecessary obstacle would I happily choose to overcome next year, just for the joy of it?

Write the answers down. They will tell you more about your sources of meaning than any forecast about AI.

9. Honest caveats

  • This is not an argument that economic disruption does not matter. Meaning is hard to find when rent is due. Jahoda's latent functions matter on top of income, not instead of it. How the gains from AI are distributed is a political and economic question this post does not try to answer.
  • The scenarios are speculative. Bostrom's solved world is a thought experiment. Amodei explicitly frames his essay as an optimistic possibility, not a prediction. Many researchers expect far slower and more uneven change.
  • Meaning is not uniform. Different cultures and individuals locate meaning in religion, family, community, craft or service in very different proportions. Any general theory should be held loosely.

Why this matters

  • The question has become practical. Capable AI makes Keynes's "permanent problem" a live issue for knowledge workers, not only a philosopher's puzzle.
  • Work gives more than income. Structure, contact, shared purpose, identity and activity need sources, with or without paid work.
  • Meaning is more robust than usefulness. Relationships, craft for its own sake and how we meet our circumstances do not depend on being the best or being needed.
  • Machines can raise human play. Go players got better and more creative after AlphaGo. Machine superiority ends only the meanings that depended on being the best.
  • Reflection is a skill. Knowing where your meaning comes from makes you more resilient to any change, technological or otherwise.

Final Thoughts

It is striking that the people building the most capable AI systems and the philosophers worrying about them have arrived, this year, at an answer Aristotle and Frankl would recognise. Meaning comes from relationships, from doing things well because they are worth doing, and from the stance we take toward our lives. None of those can be automated, because their value lies in the fact that we are the ones doing them.

The machines will keep getting better at many things. The question for the new year is not whether you can still beat them, but what you would choose to do, and with whom, if beating them were never the point.

Sources