Amazon Memo Calls AI Token Usage Metrics “Deeply Flawed”

Amazon’s newly published memo tells engineers to avoid tokenmaxxing and focus on long-term results instead of token-based AI metrics.

maisiekooc
Maisie Morrison

AgentLocker Editor

AI News
Amazon Memo Calls AI Token Usage Metrics “Deeply Flawed”

Amazon has told its software engineers to be more careful about how they use artificial intelligence tools. The guidance came in an internal memo that warns against a practice known as “tokenmaxxing.”

The memo first went out to employees in September. Amazon posted it to its website on Tuesday.

It changes how the company defines the roles of software development engineers and managers now that AI tools are part of daily work.

“We’re just seeing the beginning of AI’s effect on software engineering,” the company wrote. It added that AI “will be at the center of software engineering practice” no matter what future best practices look like.

What Is Tokenmaxxing?

Tokenmaxxing is when developers use AI as much as possible without much concern for what it produces. Tokens are the units of data that AI models process and generate.

Companies have been dealing with rising AI costs as workers adopt these tools. Some have limited how many tokens employees can use. Others have moved to cheaper, open-source models for certain tasks.

An Amazon spokesperson said the company has not placed caps on token usage. However, Amazon shut down an internal leaderboard in May after it found employees were tokenmaxxing to move up the rankings.


Amazon Shifts Focus to Long-Term Results

In the memo, Amazon said token usage, AI adoption rates, and code changes or reviews are “deeply flawed” ways to measure AI use. The company said these metrics can lead to higher costs and too much complexity.

Amazon did not tell staff to ignore those numbers completely. It said they should be weighed “with care and judgment” alongside other data.

“We believe that it is best to measure, and be measured by, long-term outcomes,” the company wrote.

Marc Brooker, a vice president and distinguished engineer at Amazon Web Services, told CNBC the company looks at “feature velocity.” This tracks how many launches teams complete and how many successful changes they make in production.

“Broadly, we think about this as a return on investment,” Brooker said. He explained that Amazon will put tools, time, tokens, and human effort into problems that matter to customers and the business.

He also said the company wants to avoid spending tokens on problems that do not matter. He noted this can be hard to judge, so some experimentation is needed.

Under CEO Andy Jassy, Amazon has quickly added AI across its business. Jassy has said AI will “change the scope and speed” of work and could shrink the corporate workforce in the next few years, though he has said it will also create “other jobs.”

The memo also tells employees not to use AI when it does not fit a task. “If your work does not benefit from current AI tooling, that is a finding, not a failing,” Amazon wrote.

The company said an engineer who can explain where AI does not yet help their work “is demonstrating the judgement we are asking for.”

Amazon shares were up 0.19% in after-hours trading on Tuesday, the day the memo was posted publicly.

From our research desk
AI Jobs Automation Index
Which jobs are AI tools targeting most? We mapped 3,400+ AI tools to real job functions — with BLS employment & salary data.
Explore the index
maisiekooc

Written by

Maisie is a news writer at Agent Locker, covering the latest developments in artificial intelligence, emerging technology and the companies shaping the future.

Discover AI Agents