I guess that tokenmaxxing phase has served its purpose, of forcing people's brains to go from "i will never use AI, it's scary and new and useless" to "i must learn to use it, this is the new normal, otherwise i'm getting fired".
And now they're able to move to the phase of "let's now learn to actually use it effectively".
I am in the “I will use AI where it makes sense to make my job easier” camp. This approach was unfortunately incompatible with my SVP’s “we need to 10x AI” goal.
I found ChatGPT to be a never ending source of ideas for how to light tokens on fire as fast as possible in order to hit my target.
Kind of surprising. I had interpreted these tokenmaxxing initiatives at some of the larger companies as R&D investment, motivating employees to find the useful applications of a new technology with fairly broad applicability. Obviously some people gamed it, but I wonder if everyone gamed it or if opening the floodgates made for any useful discoveries.
If any useful optimizations were discovered I imagine they belonged to people at the bottom of the leaderboards and hence weren't noticed by tokenmaxxing chad execs too dumb to have an ounce of creative thinking.
> What happened next was predictable in hindsight. Employees began inflating their scores through tokenmaxxing: running meaningless tasks through AI agents to consume tokens and climb the rankings.
I guess that tokenmaxxing phase has served its purpose, of forcing people's brains to go from "i will never use AI, it's scary and new and useless" to "i must learn to use it, this is the new normal, otherwise i'm getting fired".
And now they're able to move to the phase of "let's now learn to actually use it effectively".
I am in the “I will use AI where it makes sense to make my job easier” camp. This approach was unfortunately incompatible with my SVP’s “we need to 10x AI” goal.
I found ChatGPT to be a never ending source of ideas for how to light tokens on fire as fast as possible in order to hit my target.
Kind of surprising. I had interpreted these tokenmaxxing initiatives at some of the larger companies as R&D investment, motivating employees to find the useful applications of a new technology with fairly broad applicability. Obviously some people gamed it, but I wonder if everyone gamed it or if opening the floodgates made for any useful discoveries.
If any useful optimizations were discovered I imagine they belonged to people at the bottom of the leaderboards and hence weren't noticed by tokenmaxxing chad execs too dumb to have an ounce of creative thinking.
You still have near-infinite tokens, you just don't get awards for using a lot.
Day 2 Amazon only cares about the KPI.
Innovation does not grow out of KPIs.
No but often earnings do
> What happened next was predictable in hindsight. Employees began inflating their scores through tokenmaxxing: running meaningless tasks through AI agents to consume tokens and climb the rankings.
That was predictable in foresight.
Right? Make any metric meaningful for an employee, and they will find ways to maximize that metric. I thought this was a well-understand phenomenon.
It's a classic example of giving people the wrong incentive
"normalised deployments," sounds even worse than tokenmaxxing
A metric was introduced. A metric was maximized. The metric stopped being a proxy for anything. A new metric is to be introduced and likewise gamed.
Discussion: https://news.ycombinator.com/item?id=48315583