Samsung Electronics prioritizes cost-efficiency over innovation in AI transition: report

Samsung Electronics is reportedly shifting its focus in its company-wide Artificial Intelligence (AI) transition, with internal documents revealing a prioritization of cost-effectiveness over innovation. The company is introducing 'token cost-effectiveness' as a key performance indicator for employees utilizing external generative AI tools.

“

We will consider whether the tokens invested in the development and utilization of AI agents are used efficiently.

This new metric aims to measure the efficiency of AI token usage, which are the basic units of text that generative AI processes. The company's internal explanation suggests that teams achieving the same output with significantly lower AI token costs will be evaluated more favorably. For instance, a scenario is presented where two teams achieve 10 FTE (Full-Time Equivalent) results, but one uses $10,000 in tokens while the other uses $1,000; the latter group would receive higher performance marks.

This approach has led to internal concerns that Samsung Electronics might be viewing AI primarily as a tool for cost reduction rather than a catalyst for business innovation. Critics argue that metrics like FTE savings and token cost-effectiveness highlight reduced labor and AI expenses, potentially overshadowing the assessment of AI's impact on productivity, quality, and genuine innovation.

“

Token cost-effectiveness is for confirming improvements in productivity relative to investment.

Company officials, however, state that token cost-effectiveness is intended to measure productivity improvements relative to investment and to help design workforce strategies that allow employees to concentrate on high-value tasks. The move also appears influenced by the rising costs associated with AI services, as many providers are shifting from subscription models to usage-based pricing, increasing the financial burden on companies heavily reliant on external generative AI.

“

FTE represents reduced labor, and token cost-effectiveness represents reduced AI costs.

Despite the company's justifications, some internal sources express apprehension, suggesting that assessing productivity and quality improvements should precede the implementation of cost-related metrics. They believe that focusing on cost savings might hinder the exploration and adoption of AI for transformative innovation.

“

It is not too late to apply labor and cost-related indicators later, after first assessing productivity, quality, and innovation effects after AI adoption.