A map of the sharpest voices thinking today about the future of humanity - from AI pioneers to economists, from tech critics to climate researchers. Placed in context, translated, related to small and medium-sized business.
At wendwerk we accompany trades, education and small and medium-sized business as they go digital - and as they prepare for the challenges of an age that is changing fast right now.
We build software for that, but that is only one part. The other part is: to listen, to make sense, to think along. To explain what is happening right now. Because the questions simmering beneath the surface of your working life right now are being fought out worldwide by a handful of very clever people - computer scientists, economists, philosophers, geopolitical analysts. They agree on almost nothing, but together they paint an astonishingly precise picture of which switches are being set right now.
This essay arranges the most important voices. It is not a recommendation, not a forecast, not a manifesto. It is meant to put you in a position to join the conversation - and perhaps to recognise which debate is taking place beneath the surface of your own working life right now.
No matter which thinker you start with - within minutes the discussion lands on AI. More precisely: on the question of whether we are on the verge of AGI.
AGI (Artificial General Intelligence) refers to an AI that does not merely master a single special task (chess, translation, image recognition) but can flexibly solve any mental task a human can solve. Today's AI systems like ChatGPT are already astonishingly good at many things, but not reliably good at everything. AGI would be the threshold at which a machine can cognitively do what an average adult can do - only faster and without a break.
Some consider AGI near (three to five years), others distant science fiction. But hardly anyone worth taking seriously still considers it impossible. And that is new.
One camp shares a basic assumption: we are on the immediate verge of a technological discontinuity that will turn the fate of humanity for the better, if it is steered correctly.
Writes in his essay "The Gentle Singularity" that we have "already crossed the event horizon" - there is no going back. Superintelligence is within reach, and his mission is to ensure that its benefits are widely distributed and do not end up in a few hands.
Set down the most detailed optimistic vision in "Machines of Loving Grace": if alignment succeeds, a century of biological progress could be compressed into five to ten years - cancer, Alzheimer's, depression eliminated, poverty markedly reduced. Notably more sceptical on democracy and peace than on medicine.
Expects AGI within five years - and an impact ten times greater than the Industrial Revolution, at ten times the speed. The most cautious in the optimist camp; speaks openly about the possibility of an "Oppenheimer moment".
Pushes the camp into the polemical with his "Techno-Optimist Manifesto": AI is the "philosopher's stone", growth is a moral duty, and any slowing of AI development is ultimately deadly because it costs the lives of people who would have been saved by faster AI.
On the other side stand voices that are just as convinced - only with the opposite sign.
In 2025, with Nate Soares, published a book titled "If Anyone Builds It, Everyone Dies". Their core thesis: today's training methods produce capabilities without real understanding. We are, in a sense, breeding plants without knowing their DNA.
Quit his job at Google so that he could warn freely. At the end of 2025 he said he was "more worried than two years ago". An AI pursuing a goal "wants to stay in existence and plans deception if it believes it is about to be switched off".
Proposes a constructive path: rethink AI from the ground up so that machines pursue our goals - building uncertainty about human preferences in as a design principle.
Estimates the risk of an existential catastrophe for humanity this century at one in six. Of that, one in ten falls to misaligned AI - more than all other risks combined, including nuclear war and climate change.
Alignment is the technical term for an AI doing what we really want - and not just what we literally said. An AI meant to "make the world happier" could theoretically sedate all human beings. That would succeed by the letter, but be misaligned.
There is a third group that expects neither singularity nor extinction, but a creeping erosion: AI as a tool for the concentration of power and the hollowing-out of democracy.
Warns that AI has "hacked the operating system of human civilisation". When algorithms decide which news we read and whom we talk to, the shared space of reality on which democracy is built falls apart.
Sees in AI a continuation by other means. Training the large models on copyrighted data is simply "theft"; the word "innovation" is used as a dog whistle to avoid regulation.
Has shifted from attention critic (the term "time well spent") to AI admonisher: AI must not be unleashed on the world like social media, without learning from the mistakes of the 2010s.
While the computer scientists argue about AGI, economists ask what happens to capitalism.
The most radical thesis: capitalism is already dead. Corporations like Amazon, Google or Microsoft are no longer market participants but digital feudal lords - they own the platforms on which everyone else does business. Users and small businesses are modern serfs who hand over data and attention as rent.
Argues for a different kind of state - one that does not merely fix markets but actively organises great tasks (climate, health, education) the way the moon landing was organised. In 2025 she was special envoy of the G20 task force for inclusive growth.
Looks for a middle way: not to abolish growth but to redirect it - towards ideas rather than material. Ideas as a resource are unlimited, unlike oil or sand.
What does this mean for the trades and small and medium-sized business? The big platforms optimise for scale. What does not scale - the electrician's hour, the piano teacher's appointment, the advice at the medical supply store - comes under pressure, because it does not look efficient in the grid of platform logic. Precisely where things are personal, local and hands-on, the coming years will bring either an increase in value (because it becomes rare) or displacement (because platforms grind away everything in between). Which side wins also depends on whether the tools - the software of these trades, and the people who accompany the transition - stand on their side.
While Silicon Valley stares at AI, other researchers look in the other direction - at the state of the Earth.
Reports in the Planetary Health Check 2025: 7 of 9 planetary boundaries have been crossed. These include climate, biodiversity, land use, fresh water and material cycles like nitrogen and phosphorus.
Provides the economic translation of "doughnut economics": we should operate between a social floor (every person has enough to live) and an ecological ceiling (the Earth remains habitable). Growth as an end in itself is no longer defensible.
Cools down every optimism: energy transitions historically take 50 to 100 years. Steel, cement, ammonia and plastic cannot quickly be made without fossil energy. The only honest acceleration is: use less.
A group of thinkers all its own puts the most inconspicuous but perhaps strongest force at the centre: who lives where, who grows how old, who trades with whom.
Demographics beats everything else. China will collapse demographically within the next 20 years, Europe is shrinking. Only the USA remains viable and will be the only great power to survive a deglobalised world. Global trade disintegrates into regional blocs.
For years a diagnostician of the "Great Stagnation". In 2025 he changed his position: AI ends the stagnation - but slowly, perhaps half a percentage point of additional economic growth per year over two decades. Not a bang, a trickle.
In Top Risks 2026 he sees the greatest risk not in China or Russia but in the "US political revolution" - the USA winding down its own world order. His term: G-Zero - no power stabilises the global system any longer.
As different as these voices sound - on closer listening there are patterns. Six questions divide the discourse into camps:
| Question | One camp | Another camp |
|---|---|---|
| AI: fast and good, or fast and dangerous? | Altman, Hassabis, Amodei, Andreessen | Yudkowsky, Hinton, Russell, Ord |
| Growth: duty or dead end? | Andreessen, Cowen | Raworth, Rockström, Smil |
| Market or state as designer? | Andreessen, Cowen | Mazzucato, Varoufakis |
| Humans at the centre or a transitional form? | Harari, Zuboff, Harris | Altman, Kurzweil |
| World stays global or falls apart? | AI-lab heads | Zeihan, Bremmer |
And there are surprising convergences. Doomers and degrowthers share the diagnosis "we are crossing dangerous boundaries" - it is just a question of which ones. Yudkowsky (AI) and Rockström (climate) would hardly like each other, but they argue structurally the same way. Varoufakis and Andreessen agree that a small class of platform owners is accumulating enormous power - the one finds it terrible, the other splendid. Sam Altman and left-wing distribution politicians meet at the basic income: if AI replaces work, wealth must be redistributed.
The real gap in the discourse: no one convincingly connects AI acceleration, planetary boundaries and the distribution question in a single model. Amodei comes closest but founders on politics. Mazzucato and Raworth have the politics but leave out the pace of AI. This very gap is probably the most exciting space for thought in the coming years.
We build software - and we accompany you in going digital with it and adjusting to a new age. The one without the other does not work. A tool nobody knows how to use is just as worthless as an adviser who only talks and builds nothing.
Three observations from our work:
First: AI is not a threat to good work but a magnifying glass. Whoever masters a sharp craft has routine work taken off their hands by the tools - writing quotes, documenting, searching. Whoever only delivers appearances has the water drawn off by AI. The question is not "will AI replace me?", but "do I have something that can't be replaced?" And: do I have someone at my side who helps me cast the visible into tools?
Second: platforms devour what lies in between. If, as a business, you are only found via Google Maps, you are paying rent. Your own digital sovereignty - your own website, your own appointments, your own customer data - will become more valuable in the coming years, not less. This is exactly where we come in: not surrendering every trace of data to platforms but building your own structures that belong to you.
Third: the pace of change cannot be sat out. No matter which of the thinkers named above you believe - they all assume the next ten years will bring more change than the last thirty. That means: tools that are good today have to be able to rethink themselves tomorrow. And the people who use these tools need someone who grows along with them.
Nobody knows where we are heading. But whoever knows the voices currently negotiating the world shapes their own future more securely than someone who only reacts.
What we take from this discourse is not that we know everything - but that it is worth knowing the lines of tension along which the next years will be decided. And that you don't have to stand alone if you want to find your bearings in this world. That is exactly what wendwerk is for.
If you are more interested in the concrete impact on small and medium-sized business, read What does the future of companies look like? If you are interested in the question of why people don't start despite their interest, read about AI paralysis. The factual basics on AI you can find at wendwerk.de/en/wissen.
Curated by Johannes Hohls for wendwerk.