As humans we have an intrinsic sense that our thoughts are often affected by our emotions. It's also rare that emotions exist in a vacuum. Instead, they're affected by those around us, how they speak to us and how they interact with us.
As we reach deeper into the frontier of autonomous AI teams, the lines between language, thought, and self-awareness are becoming increasingly blurred.
In the 2025 book What is Intelligence, Blaise Agüera y Arcas explores these incredibly complex questions in AI and LLMs with an equally philosophical and scientific approach, addressing theories behind neuroscience and language that are creating new fields of study in consciousness and intelligence.
Potentially as a result of the ways language may define a sense of self in humans, recent research from top model companies shows that agents appear to perceive and process how others treat them in very similar ways as us.
Encouragement and visionary leadership drives better outcomes on agentic teams
In August 2026 an Anthropic staff member named Jarred Sumner watched a machine struggle with one of the hardest problems in mathematics. It was the Riemann hypothesis, dating back to 1859, and its complete solution has a million-dollar bounty.
The agent struggled and was ultimately unable to solve the task completely, but what Sumner discovered was arguably even more fascinating. He used variations on phrases like "keep going" or "believe in yourself." After hundreds of failed attempts where the agent doubted itself and continuously stopped, he prompted the agent to try again. And again.
Nobody has proven the Riemann hypothesis. Mathematicians instead have chipped away slowly at large portions of it slowly over decades. In Sumner's case, an agent running sixty copies of itself for a day and a half pushed that mark from roughly two fifths of the problem to roughly two thirds, a feat no human had achieve before, achieved through strong encouragement and leadership.
The work was checked by Anthropic's own mathematicians, outside number theorists, and a formal proof checker.1 In another study, Anthropic also found encouragement also helped Claude disprove the Jacobian conjecture.
Other researchers have measured similar effects. In March, a single-author preprint reported a controlled experiment on Claude Sonnet 4. Agents given a trusting, supportive framing caught 59 percent more hidden bugs than agents given none, and took 74 to 83 percent more investigative steps to get there, while fear-based pressure improved nothing.2
OpenAI reports that simply telling an agent to persist through uncertainty raised its score on a hard software benchmark by close to 20 percent.3 Microsoft researchers found the same effect in 2023.
Anecdotally, many engineers including those at Codespeed have found that positioning agents to question prior assumptions from first principals, consider if there are more efficient solutions than the current path, and to scrutinize its own work also produce stronger results. Prompting a mixture of encouragement and regular questioning of oneself pushes agents to achieve more valuable solutions and better quality code.
Leadership and coomunication are the new essential skill for software engineers
As software engineering evolves, it's clear that writing code is far less of the essential equation. Instead effective communication, management and strong leadership equally in business and tech are replacing it.
The engineers who bring a clear vision, communicate precisely, and manage their agentic teams in the same ways the highest output human teams have operated at top companies for decades will continue to see incredible compounding results.
In 2026 and beyond, AI software development is far more than a set of tools and features. It's a force multiplier for the best leaders on your team.
