Top Indeed exec details why they'll never have a 'Tokenmaxxing'-esque leaderboard

Indeed is carefully monitoring the use of artificial intelligence among its employees, but it is steering clear of implementing any form of leaderboard akin to the controversial 'Tokenmaxxing' model. In an interview with Business Insider, Anthony Moisant, the company's chief information officer, emphasized that while they do track token usage, it remains a behind-the-scenes focus. 'We definitely are not going to use a leaderboard,' he stated. Moisant acknowledged that there isn't anything fundamentally wrong with the leaderboards used by Meta and other major tech firms to incentivize AI token usage. However, he believes that such metrics can lead to distorted incentives, where employees may pursue token accumulation rather than meaningful outcomes. 'Anytime we have a metric that becomes part of an incentive system, we risk creating a perverse incentive that can lead to unintended consequences,' he explained. The tech industry has been pushing for greater employee engagement with AI tools. Nvidia's CEO, Jensen Huang, recently expressed expectations that high-earning engineers should utilize a substantial amount of AI resources. Tokens, which are essential for large language models, serve as the numerical foundation for AI systems like OpenAI's ChatGPT. Instead of concentrating on token consumption, Indeed prioritizes metrics that reflect product delivery speed and customer feedback. 'Activity does not equate to value for our business or the job market,' Moisant remarked. 'What truly matters are the outcomes, such as improved and faster matching processes.' Moisant cautioned against placing excessive emphasis on easily measurable metrics, recalling a past initiative where Indeed tracked the percentage of code written. Although it seemed like a beneficial measure initially, it proved inadequate in driving genuine productivity improvements. The issue of 'Tokenmaxxing' also highlights the increasing financial commitment many companies are making towards AI technology. Moisant noted that Indeed's AI-related expenses are anticipated to quadruple by 2025, largely due to research and development efforts. The board recently engaged in discussions about maintaining budgetary discipline while still embracing the advancements that AI tools can offer. 'We had a robust debate about our spending priorities,' Moisant shared. 'While we need to control costs, we also don't want to hinder our productivity and the outcomes we're aiming for.' He reiterated the importance of being mindful of expenditures, even in high-return areas where spending is generally encouraged. 'For example, we are keen on monitoring how quickly we can translate ideas into products for customers,' he added. 'Understanding our spending patterns and identifying spikes in token use can help us gauge where the value lies.'