The advancement of AI is not linear. While the industry anticipated a "year of agents" for practical assistance, the most significant recent progress has been in specialized, academic fields like competitive mathematics. This highlights the unpredictable nature of AI development.

People deeply involved in AI perceive its current capabilities as world-changing, while the general public, using free or basic tools, remains largely unaware of the imminent, profound disruption to knowledge work.

As AI achieves impressive milestones, like assisting in creating a cancer vaccine, the public conversation immediately discounts the achievement. The goalposts shift from "AI helped solve a problem" to demanding a fully autonomous, one-shot solution. This pattern of escalating expectations obscures the real, incremental progress being made.

AI models will produce a few stunning, one-off results in fields like materials science. These isolated successes will trigger an overstated hype cycle proclaiming 'science is solved,' masking the longer, more understated trend of AI's true, profound, and incremental impact on scientific discovery.

Non-tech professionals often judge AI by obsolete limitations like six-fingered images or knowledge cutoffs. They don't realize they already consume sophisticated AI content daily, creating a significant perception gap between the technology's actual capabilities and its public reputation.

A major frontier for AI in science is developing 'taste'—the human ability to discern not just if a research question is solvable, but if it is genuinely interesting and impactful. Models currently struggle to differentiate an exciting result from a boring one.

Now that AI can churn out a competent, human-level research paper daily, the incentive for incremental work disappears. To stand out, the scientific community must leverage AI as a tool to raise its ambitions and tackle grander, more fundamental problems.

Bret Taylor explains the perception that AI progress has stalled. While improvements for casual tasks like trip planning are marginal, the reasoning capabilities of newer models have dramatically improved for complex work like software development or proving mathematical theorems.

Frontier AI models exhibit 'jagged' capabilities, excelling at highly complex tasks like theoretical physics while failing at basic ones like counting objects. This inconsistent, non-human-like performance profile is a primary reason for polarized public and expert opinions on AI's actual utility.