Nature: Top 100 Fastest-Growing Technologies According to AI
Australian researchers used the Wikipedia2Vec model to compile the Momentum 100 ranking, where reinforcement learning, blockchain, and 3D printing are named the fastest-growing technologies of 2026 based on Wikipedia data analysis.
Momentum 100: How AI, for the First Time Without Experts, Identified the Fastest-Growing Technologies of 2026
Introduction
Every year, leading global publications and think tanks publish rankings of the most promising technologies. MIT Technology Review, Stanford University, the World Economic Forum — all rely on the opinions of a narrow circle of experts, whose judgments, however authoritative, are inevitably subjective.
In April 2026, the Australian analytics company League of Scholars proposed a radically different approach. In collaboration with Nature Index, it presented the Momentum 100 ranking — the first-ever list of the most dynamically developing technologies, compiled not by humans but by artificial intelligence based on Wikipedia analysis. AI examined tens of thousands of articles, analyzed millions of hyperlinks and page view dynamics to determine, without any human intervention, which technologies are gaining momentum the fastest.
Event Details and Timeline
The Momentum 100 ranking is based on the open dataset Cosmos 1.0, published in the journal Scientific Data, part of the Nature group. To create it, the League of Scholars team used the Wikipedia2Vec language model, which converts Wikipedia articles into multidimensional numerical vectors — so-called "embeddings."
The key advantage of this method is that embeddings capture not only the content of articles but also the logic of hyperlinks between them, reflecting real semantic connections in scientific and technical discourse.
The methodology included several stages:
- Analysis "seed" — a single starting article titled "List of emerging technologies."
- Network construction — the algorithm extracted nearly 55,000 interconnected Wikipedia pages.
- Filtering — of these, over 23,000 pages directly related to technologies and related concepts were retained.
- Evaluation — each technology received an index based on age, page view dynamics, and other metrics.
Validation of results was carried out by comparison with other datasets, including academic publications and patent information.
The creation timeline of Cosmos 1.0 is described as a labor-intensive process that should have taken six months, but the League of Scholars team, with support from programmers and designers, completed it in one week of non-stop work.
Impact and Significance
For science and analytics, the emergence of Momentum 100 means a paradigm shift in forecasting methods. As noted by Catherine Aiken from Georgetown University, who specializes in emerging technologies, "over the past six years, methods for identifying promising areas have hardly been updated — they are too expert-oriented, labor-intensive, and individualized." She called Cosmos 1.0 a "useful addition" to this field, opening up opportunities for more objective analysis.
The top 10 technologies of the Momentum 100 ranking (first three positions: reinforcement learning, blockchain, and 3D printing) also include soft robotics, augmented reality, and omics technologies — large-scale studies of biomolecules such as DNA, proteins, and metabolites.
Key technologies of the ranking:
| Rank | Technology | Key Characteristic |
|------|------------|-------------------|
| 1 | Reinforcement Learning | The system learns through trial and error, receiving a "reward" for correct decisions |
| 2 | Blockchain | Applications extend far beyond cryptocurrencies: medical data, supply chains, energy |
| 3 | 3D printing | Additive technologies continue their expansion in industry and medicine |
Why is reinforcement learning in first place?
The versatility of the method, which allows AI to make sequential decisions in a complex, constantly changing environment, secured its leadership. AI based on reinforcement learning already beats champions in Go and chess, is used in drug development and drone racing. The ranking creators note that the algorithm mathematically replicates natural learning mechanisms — roughly how animals learn commands by receiving treats.
Blockchain: from cryptocurrencies to swarm learning
Interest in blockchain, which placed it second, is fueled not by Bitcoin but by publications in leading scientific journals. One article, which received over 800 citations, describes swarm learning — a method that allows hospitals and laboratories to jointly train AI on medical data without disclosing patients' personal information.
Reaction of Key Players
Paul McCarthy himself, co-founder of League of Scholars, defines the project's philosophy as an attempt to "map technologies from the bottom up," using AI's ability to discover hidden knowledge in large complex systems.
Nature, which published the article about the ranking, thereby legitimizes this approach before an academic audience. The journal's authority serves as an important signal to the scientific community: AI data can be as reliable as expert assessments.
Catherine Aiken from Georgetown University, whose assessment is cited both in Nature and on a Ukrainian popular science portal, expresses cautious optimism. She calls Cosmos 1.0 a "useful addition" to existing methods but does not advocate completely abandoning the expert approach.
Forecast and Conclusions
Short-term forecast (2026-2027): League of Scholars plans to make the ranking annual, tracking changes in dynamics. This will allow not only recording current leaders but also identifying technologies that are losing or gaining momentum.
Long-term forecast (2028+): If the approach proves effective, we may see the spread of similar methods in corporate strategy, venture investment, and government planning. AI analysis of large volumes of unstructured data (scientific articles, patents, news feeds) can identify technological trends earlier than even the most attentive experts notice them.
Main conclusion: Momentum 100 is not just another ranking. It is a demonstration of how AI can take over functions traditionally considered the prerogative of human experts. Analyzing 55,000 Wikipedia articles and identifying hidden connections between them is a task impossible for any group of analysts but trivial for a well-trained model.
It is telling that the #1 technology in the ranking is the very method that made it possible to compile it. Reinforcement learning is the same "trial-error-reward" paradigm that nature uses for everything — from dog training to species evolution. And the AI that compiled the ranking essentially applied the same principle to the analysis of human knowledge. It turned out to be a kind of recursion, which is perhaps the best proof that the choice of technology is absolutely accurate.
— Editorial Team
No comments yet.