What 81,000 People Want from AI: The Business Owner's Guide to User Expectations

In March 2026, Anthropic published the largest and most multilingual qualitative study of its kind. Nearly 81,000 people participated. The study asked a simple question: what do you actually want from AI?

The results are useful for business owners designing AI-powered customer experiences. Not because they reveal surprising preferences, but because they confirm what thoughtful practitioners already suspected. Users want AI to help with specific tasks, respect their boundaries, and stay out of areas that require human judgment.

Here are the key findings and what they mean for businesses building AI experiences.

Top Five Things Users Want AI to Help With

The study identified clear patterns in user requests.

1. Information synthesis and research. Users want AI to gather, summarize, and explain information. Not to generate new content from scratch, but to make existing information more accessible.

2. Writing and editing assistance. Drafting emails, reports, and documents. Refining existing text. This is AI as a tool, not a replacement.

3. Technical problem-solving. Debugging code, explaining technical concepts, troubleshooting. Users treat AI as a knowledgeable colleague.

4. Learning and education. Understanding new topics, getting explanations tailored to their level. AI as a tutor.

5. Productivity and organization. Managing tasks, scheduling, prioritizing. AI as an executive assistant.

What these have in common: they're all enhancement tasks. Users want AI to make them more capable, not to take over completely.

Top Five Fears Users Have About AI

The study also revealed consistent concerns.

1. Job displacement. Users worry AI will replace their work entirely, not augment it.

2. Privacy and data security. What happens to the information they share with AI systems?

3. Accuracy and reliability. Users don't trust AI to be correct, especially for important decisions.

4. Loss of human connection. AI interactions feel transactional. Users miss the nuance of human conversation.

5. Over-dependence. Users worry about losing skills if they rely too heavily on AI assistance.

These fears aren't irrational. They're grounded in real limitations of current AI systems.

Implications for Customer-Facing AI Experiences

If you're building AI into your products or services, these findings have practical implications.

Position AI as an enhancement, not a replacement

Users respond better to AI when it clearly augments their capabilities rather than replacing their judgment. A customer service AI that helps customers find information is more accepted than one that tries to resolve disputes independently.

Be transparent about limitations

Users don't expect perfection. They expect honesty. When AI systems acknowledge uncertainty or recommend human review for certain decisions, users trust them more.

Design for the right tasks

The study shows users want AI for information synthesis, writing assistance, and technical problem-solving. If you're deploying AI for emotional support or complex decision-making, expect resistance.

Address privacy explicitly

Users care about data security. Your AI experiences should explain what data is collected, how it's used, and what controls users have.

Maintain human touchpoints

Users value human connection. Even heavily automated experiences should have clear paths to human interaction when users want them.

How to Communicate AI Use Without Triggering Distrust

The study suggests communication strategies that build trust.

Lead with capability, not intelligence. Users trust AI that does specific things well. "Our AI helps you find relevant documents" is better than "Our AI understands your needs."

Acknowledge limitations upfront. "This AI assistant can help with common questions. For complex issues, you'll be connected to a human specialist."

Give users control. Let users choose when to use AI and when to opt for human interaction. Forced AI experiences feel manipulative.

Be specific about data practices. Generic privacy policies feel evasive. Explain what you actually do with user data.

Design Principles for AI Experiences

Based on these findings, here are design principles for business owners.

1. Start with enhancement tasks. Information synthesis, writing assistance, technical explanation. These align with what users want.

2. Keep humans in the loop for judgment tasks. Decisions, disputes, emotional situations. Users want human involvement.

3. Be explicit about scope. Tell users what the AI can and cannot do. Uncertainty breeds distrust.

4. Provide human alternatives. Users should always be able to reach a person if they prefer.

5. Explain data practices clearly. Don't bury privacy information. Make it accessible and understandable.

For more on designing customer experiences that align with user expectations, see our guide on AI customer journey mapping. And if you're implementing AI-powered support, explore our SnappyClaw solution.

The Bottom Line

81,000 users told Anthropic what they want from AI. They want it to help with tasks, respect their boundaries, and stay out of areas that require human judgment.

This isn't rocket science. It's common sense that gets lost in the excitement about AI capabilities.

The businesses that succeed with AI aren't the ones with the most sophisticated technology. They're the ones that deploy AI in ways that align with what users actually want. Enhancement, not replacement. Transparency, not black-box automation. Human touchpoints, not forced AI interactions.

Listen to what users are telling you. They're usually right.