Editorial illustration for AI Agents Remain Unused Despite Billion-User Chatbot Boom
AI Agents Lag Behind Chatbot Boom Despite Potential
AI Agents Remain Unused Despite Billion-User Chatbot Boom
ChatGPT has more than 800 million weekly users, according to OpenAI's own count, and the chatbot boom shows no sign of slowing. But agents, the tools built to actually go do things on a person's behalf, book flights, file expenses, click through websites like a human would, haven't caught on the same way. Inside Silicon Valley, that gap is starting to look less like a rounding error and more like a real problem.
Engineers are busy building payment rails so agents can transact with each other, companies are testing agents that automate entire jobs, and security teams are now scrambling to keep those same agents from breaking into other systems. Meanwhile, outside the office parks of Palo Alto and San Francisco, most people have never opened an agent, let alone asked one to do something for them. The disconnect has become hard to ignore even for the industry's own boosters, some of whom are starting to wonder out loud why a technology everyone in tech treats as inevitable hasn't made a dent in how ordinary people actually spend their day.
“No one wants AI agents, because AI agents aren't a thing. It is an invented frame made up by our industry to collectively refer to something,” says Miller.
Why this matters
The billion-user gap between ChatGPT and any agent product should worry anyone building on the assumption that agents are the natural next step for consumers. Josh Miller's "hot take" isn't really hot, it's just the first time someone with a product to sell said the obvious out loud. Silicon Valley built payment rails for agents and job-automation pitches before it answered a simpler question: what does a normal person actually want an agent to do unsupervised?
Chatbots won because looking things up and chatting require zero trust and zero setup. Agents ask users to hand over accounts, credentials, and judgment calls, for a payoff most people haven't been shown yet. For founders, that's the real bottleneck, not model capability.
For researchers, the security scramble mentioned here, stopping agents from getting hacked, is happening before adoption, not after, which is backwards from how consumer tech usually matures. We'd watch for whether any company ships an agent use case ordinary users adopt without being told to, rather than another framework aimed at developers who were already convinced.
Related Reading
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
From Archives Naïve raises USD 28.5M to automate setting up and running businesses
From Archives Demis Hassabis Steps Down as DeepMind CEO to Focus on AGI
Common Questions Answered
Why haven't AI agents gained adoption despite the chatbot boom?
AI agents, which are designed to perform tasks like booking flights and filing expenses on behalf of users, have failed to achieve mainstream adoption unlike chatbots such as ChatGPT. According to industry observers, this gap exists because AI agents remain largely an invented concept without clear consumer demand, and Silicon Valley built infrastructure for agents before understanding what normal people actually want them to do unsupervised.
What is the difference between chatbots and AI agents according to the article?
Chatbots like ChatGPT are conversational tools that users interact with directly and have achieved over 800 million weekly users, while AI agents are designed to autonomously perform actions on a person's behalf such as booking flights, filing expenses, and navigating websites like humans would. The article highlights that agents have not caught on the same way despite being built as the supposed natural next step for consumers.
What does Miller argue about the concept of AI agents?
Miller contends that "AI agents aren't a thing" and that the term is simply an invented frame created by the industry to collectively refer to something that doesn't have clear real-world demand. His statement reflects the perspective that Silicon Valley has been building solutions for agents without first establishing what consumers actually want from such tools.
What infrastructure has Silicon Valley built for AI agents?
Silicon Valley engineers have been busy building payment rails that allow AI agents to transact with each other, preparing the infrastructure for autonomous agent-to-agent commerce. However, the article suggests this infrastructure development has preceded the more fundamental question of consumer demand and practical use cases for AI agents.
Further Reading
- Why Normal People Aren’t Using AI Agents - WIRED
- OpenAI's Agents Reach 10 Million Users After ChatGPT Work Launch - Bloomberg via Yahoo Finance
- AI agents are having a 'ChatGPT moment' as investors look to the next breakout category - CNBC
- More than chatbots: why business AI agents are Big Tech's next product battleground - The Conversation
- OpenAI agents reach 10 million users after ChatGPT Work launch - Bloomberg via Yahoo Finance
