Sam Altman criticizes token pricing and unveils Dots at OpenAI DevDay

Beyond the Industrial Revolution Paradigm

President Donald Trump often characterizes artificial intelligence as an Industrial Revolution on a massive scale. Altman rejected that specific historical framing during his conference remarks, arguing instead for a different lens through which to evaluate technological impact.

Altman asserted that artificial intelligence will wield an even greater influence on human history than the Industrial Revolution did. Yet he cautioned against viewing the technology merely as a tool to accelerate economic output without broader human empowerment.

In Altman’s view, the overarching goal of this technology should not simply be to crank up economic productivity merely to benefit artificial intelligence systems. I think we should learn from the mistakes of the Industrial Revolution, and I think we should be more like the Renaissance.

The Problem with Token-Based Pricing

Tokens serve as the fundamental unit of input and output for modern language models. Companies across the technology sector utilize tokens to price their Application Programming Interface models, establishing a baseline where corporate clients pay per cluster of characters.

Some developers engage in tokenmaxxing, artificially inflating their daily output metrics by consuming maximum token volumes to signal productivity to management. Altman directly criticized this standard during the San Francisco event.

Tokens are a terrible metric for a bunch of reasons, Altman stated, noting that different underlying model architectures churn through wildly varying numbers of tokens to accomplish identical tasks. Consequently, raw token counts fail to serve as a reliable proxy for actual business value.

OpenAI product teams explored eliminating token metrics entirely, but customer resistance halted the shift. Corporate buyers demanded an underlying basis for expenditure, leaving tokens as the default billing mechanism for the immediate future. We have not figured out a better way to do that, Altman admitted to attendees.

Introducing Dots and Mitigating Digital Exhaustion

Expanding the company hardware and software ecosystem, Altman showcased Dots, a personal AI agent designed to handle routine tasks and process notification queues. Altman noted it is reasonable to assume Dots will run inside an upcoming dedicated hardware device built by OpenAI.

For the OpenAI chief executive, the agent serves a practical psychological function in his daily routine. By intercepting overnight updates and managing notifications before he wakes up, Dots helps mitigate smartphone addiction and digital fatigue.

I feel like I have finally gotten some of my attention back, Altman said. I no longer feel quite addicted to my phone in the same way. Going through everything from overnight in the morning would make my brain feel polluted, and now my Dot just deals with the stuff.

Managing the AI Safety Divide

Amid rising cyber incidents and corporate anxiety over autonomous agents going rogue, Altman addressed the polarizing debate surrounding artificial intelligence safety. He characterized current industry discourse as divided into two flawed extremes.

On one side sit proponents of rapid capability scaling who downplay safety protocols. On the other side are factions advocating for a total halt or shutdown of advanced research. Altman argued that both paths carry distinct operational risks.

I can imagine things going wrong in either direction, Altman said, citing potential loss-of-control incidents during capability sprints versus extreme power concentration if development stalls entirely. He maintained that safety measures must stay ahead of raw capability advances while the industry pursues what he termed a centrist path.

Assessing the Modern Labor Market

Addressing young professionals entering the workforce during an uncertain macroeconomic climate, Altman offered an optimistic outlook despite persistent employment hurdles. US Census Bureau and Bureau of Labor Statistics data place the unemployment rate for recent college graduates at 5.7% as of June.

Responding to an attendee inquiry during the Q&A session, Altman shared his perspective on timing by remarking that entering the workforce right now is as advantageous as it gets, apart from hopefully all future moments.

While acknowledging that specific job classifications will disappear entirely due to automation, Altman argued that broader economic growth projections outweigh dystopian predictions of permanent mass unemployment. If you are an ambitious person graduating college right now, I think it is probably the best time thus far in history, he concluded.