Every prediction about the imminent "death" of progress in artificial intelligence—from slowing learning to hitting a "wall"—has consistently proven wrong. An analysis conducted by the Rand Group team clearly demonstrates that all these "funerals" for AI shared the same problem: a lack of substance. Development has not only failed to stop but has accelerated.
The experts illustrated their thesis with a graph, where the black line represents actual progress and the yellow lines represent predictions. The key metric is the length of a task in person-hours that AI can complete in half of the cases. And this curve is steadily rising.
Loud claims shattered by reality
Two notable episodes are cited as examples. In March 2022, an article was published with the bold title "Deep Learning Hits a Wall." About a year later, the GPT-4 model emerged, fundamentally changing perceptions of AI's capabilities.
The second episode dates to November 2024. At that time, claims emerged that "the returns from pre-training models have plateaued, and the models themselves are hitting a ceiling." For an entire month, "obituaries" on this topic spread across the industry. However, while these farewell speeches were being written, the black line of progress did not slow down—it accelerated.
The trap of the "mature stance"
Why were these predictions so easy to believe? The 2024 claim sounded particularly convincing because it came from people who were themselves creating the technology. According to analysts, this is precisely where the trap lies. Predicting a plateau always seems like a "mature," skeptical, hype-free stance. Yet, it has proven to be a losing position for seven consecutive years. Every prediction of an imminent limit has been shattered by continued growth.
Expert opinion: The key takeaway from this analysis for me as a crypto market analyst is obvious: betting on a halt in progress is a systematically losing strategy. Just as some investors constantly wait for "Bitcoin to correct to zero," AI skeptics each time find new reasons why development should slow down. Reality, however, demonstrates the opposite: exponential growth in capabilities continues, and those who bet on it come out ahead in the long run. Skepticism is not synonymous with wisdom, especially when it contradicts the data.