Amazon's forced AI adoption will massively backfire
Don't burn tokens just to impress your boss
We now live in a world where you’re forced to use AI (no matter what), or you’ll lose your job.
So those who prove they can burn the most tokens, even if it doesn’t produce anything useful, get the pay raise.
This is exactly what’s happening at Amazon, and quite likely at other tech companies too.
Employers have bought into the AI hype train and insist that everyone has to use AI in every single workflow.
Scoreboards are used to track your AI usage, and employees will use them in any other way except for being productive, just to burn more tokens in this ridiculous PvP game.
This will not end well
Goodhart’s Law was something I learned when I was interested in crypto airdrops:
These were essentially free money that you get when you interact with a project before they launched their token.
And because this ‘free money glitch’ worked so well, others started using the airdrop criteria used to reward past airdrops and use it for future ones.
Some rewarded the number of transactions signed on a blockchain, so one person can spin up 1,000 wallets doing the exact same actions.
It was hoped that all of them would be seen as ‘real humans’, so they can get the tokens and dump them (they’re also known as Sybils).
This is the perfect example of Goodhart’s Law:
When a measure becomes a target, it ceases to be a good measure.
And this is happening to AI usage as well:
When companies reward token burn, employees will do whatever it takes to burn more tokens than their colleagues.
Even if they’re using it for non-productive work.
Everyone starts to optimise for token burn, which directly goes against the business goals:
Your employer needs to see a return on token spend
Any employee and any tool will only make sense for your employer if they do either of these:
Make more money
Save more money
That’s essentially how you prove your value to your company, when you tie it directly to revenue.
But now that tech companies are prioritising token spend, it’s encouraging employees to waste money instead.
This can’t go on forever.
AI subsidies will soon be removed once VCs want to profit from their investments.
Tokens will get more expensive, and employees who focus on token burn will become a cost liability.
So instead of playing this ridiculous game, here’s what I’d do to prove that I’m using AI effectively:
Build real substance with AI
Instead of building a flashy website, I’d focus on building practical workflows with AI:
What are some bottlenecks in the company that stop it from making or saving more money, and how can I use AI to remove them?
There’s no need to pretend that you’re using AI well, when you have real outcomes and proof that you are AI-native:
You automate the tasks that you hate doing so that you have more time to do the high-leverage tasks that AI can’t do for you.
And if you want to learn how to build these workflows, I’ll share what I learnt after spending 4 months with AI in The AI-Native Sprint.
This is a 90-minute crash course that gives you a clear path to becoming AI-Native for busy 9-to-5 professionals.
If you’d like to join us, sign up for the waitlist here: