Maximizing AI: More Than Just Token Counting
Synthesia's HR leader argues for focusing on AI's business impact rather than token usage. As AI costs surge, the true measure of success isn't in the tokens burned but in tangible outcomes.
In the current AI-driven landscape, the notion of 'tokenmaxxing', focusing on maximizing AI token usage, has come under scrutiny. Laura Gonzalez, head of people at Synthesia, argues that the true value of AI lies in its impact on business outcomes, not in the sheer volume of tokens consumed.
Shifting Priorities
Focusing solely on AI token usage, Gonzalez suggests, is akin to measuring a salesperson's success by the number of calls made rather than the deals closed. It's a flawed metric that many in the industry are starting to question. Synthesia, an AI video company, exemplifies this shift by not incentivizing token burn rates but instead prioritizing how AI enhances efficiency and innovation across its teams.
Companies like Coinbase have experienced a spike in AI usage, leading them to introduce spending caps ranging from $500 to $5,000 weekly based on job roles. This isn't a decision driven by cost alone. It's about redefining what AI success looks like. Marty Kausas, CEO of Pylon, echoes this sentiment, noting a substantial increase in their annual AI costs from $400,000 to $1.4 million. The industry is moving away from the era of tokenmaxxing.
The Real Metric: Business Impact
Gonzalez emphasizes that the focus should be on how AI contributes to tangible business improvements. Are teams able to deliver projects faster? Is customer interaction more effective? Are new hires making an impact sooner? These are the questions that matter. While AI usage is encouraged, it's not the sole indicator of innovation or success.
Despite discouraging rampant token usage, Synthesia maintains a culture of experimentation, providing employees the tools to explore AI's potential without excessive token incentives. This approach suggests that innovation doesn't require burning through AI resources at unsustainable rates.
Token Usage: When Does It Matter?
While token use isn't the primary focus, there are moments, such as hackathons, when it's beneficial to track AI activity. But these are the exceptions, not the rule. For new hires, understanding AI resource access is key, much like asking if a job includes a laptop in the past. It's about ensuring they've the necessary tools to succeed.
The AI-AI Venn diagram is getting thicker as companies realize that AI's true value lies in its business impact, not in token usage. The question remains: Are we measuring what truly matters?
Get AI news in your inbox
Daily digest of what matters in AI.