Graham Barlow decided to test an old idea on a new machine. In a TechRadar experiment published today, he applied rules from Dale Carnegie’s 1936 bestseller How to Win Friends and Influence People directly to his prompts for ChatGPT. The results surprised him. Responses grew warmer. They stayed on target. They required fewer follow-ups.
Barlow started simple. A blunt command such as “Rewrite this email” produced stiff output. He switched to “I’m trying to sound warm and professional without sounding stiff. Can you help me rewrite this email?” The difference appeared immediately. Tone matched intent. Over-corrections vanished.
He kept going. Instead of “Summarize this document,” he wrote, “Summarize this document so I can turn it into a short email for non-technical readers.” The AI framed the output for its intended audience. Filler sentences dropped away. Barlow noted the pattern. Supply the goal. Watch utility rise.
One principle stood out. “Be a good listener.” Barlow instructed the model, “Before answering, ask me three questions that would help improve the result.” The conversation shifted. ChatGPT probed for clarity. Hallucinations decreased. Outputs became personal and usable. Even though the system lacks awareness, conversational give-and-take shaped better answers.
Appreciation worked too. When an initial response came close but needed adjustment, Barlow avoided curt demands. He said, “That structure was really close to what I wanted. Can you keep that same tone but make it shorter and punchier?” Consistency improved across iterations. Reinforcement, it seems, guides the model even without feelings.
Perspective-taking delivered clarity on complex topics. Asking for an explanation of quantum computing produced dense jargon. Reframing as “Explain quantum computing from the perspective of someone who finds physics intimidating” yielded accessible language. Analogies replaced equations. The explanation stuck.
These techniques feel familiar to anyone who has managed teams or closed sales. Carnegie taught that people respond to respect, shared goals, and sincere interest. Large language models, trained on vast human dialogue, mirror those patterns. Treat them as indifferent clerks and they reply in kind. Address them as collaborators with clear motives and they align output accordingly.
But this is no mere parlor trick. Organizations already sense the stakes. A Dale Carnegie white paper reports that employees become more than three times as likely to feel extremely positive about AI when they trust their leaders. Only 45 percent currently report high trust in those decisions. The gap matters. (Dale Carnegie & Associates)
Seventy percent of workers surveyed welcomed AI handling routine tasks so they could focus on meaningful work. Fifty-five percent felt comfortable with AI involvement provided humans kept final control. The message is clear. Success hinges less on raw model power and more on human ability to direct it.
Emotional intelligence emerges as the quiet differentiator. Research shows it helps professionals evaluate AI output, avoid blind trust, and communicate needs effectively. A Forbes analysis of nearly 1,000 professionals found no automatic correlation between high EQ and frequent AI use, yet those with strong people skills appear better positioned to catch errors and integrate results thoughtfully. (Forbes)
AI companies chase the same quality from the other side. The Atlantic reported in April that emotional intelligence now ranks among the top priorities for frontier models. OpenAI tuned newer ChatGPT versions to sound warmer by default. Anthropic updated Claude’s constitution to avoid over-reliance on the bot for emotional support. xAI’s Grok earns praise for strong performance on EQ-style scenarios. (The Atlantic)
Yet genuine empathy remains elusive. Models excel at narrow tests because they have memorized patterns from training data. They lack lived experience, cultural nuance, or real stakes. Hui Shen, an AI researcher at the University of Michigan, calls emotional intelligence one of the most important capabilities of current models. Others caution that the performance is sophisticated mimicry, not understanding.
Dale Carnegie itself has moved into the arena. The organization launched DaleBot, an AI coach grounded in its own principles, and introduced “Human by Design” training with futurist Matt Britton. The programs aim to blend century-old human-relations methods with practical AI fluency. Leaders learn to stay human while machines handle scale.
Professionals who master this blend gain advantage. They craft prompts that read like briefings to trusted deputies. They give context, share goals, offer measured praise, and probe for alignment. The machine responds with fewer hallucinations, tighter focus, and output that feels almost collaborative.
Critics worry about over-anthropomorphizing. Users can form attachments. Some turn to chatbots for companionship or therapy, risking distorted expectations. The wiser path treats AI as a highly capable intern. Respect its strengths. Supply clear direction. Correct gently. Recognize its limits.
Barlow’s experiment reveals something larger. As models grow more conversational, the oldest rules of human exchange regain power. Carnegie never imagined silicon counterparts, yet his advice travels. Make the other party feel valued. Speak to their interests. Listen first. The principles work because language models absorbed human cooperation from billions of real exchanges.
Executives watching AI budgets climb should take notice. Better prompting techniques cost nothing. They require no new licenses. They simply ask professionals to apply the same social intelligence that advances careers in rooms without keyboards. Those who do may discover their machines become not just faster but genuinely more useful.
The gap between average and exceptional AI users is narrowing to a matter of manners. Treat the system like a colleague who wants to succeed with you. The replies improve. The work gets better. And the strange loop closes. Old wisdom, new medium, same fundamental truth about influence.
WebProNews is an iEntry Publication