Urgent.News

What's breaking now, across thousands of outlets.

AI

Most work doesn’t need the Ferrari of AI, Wharton and Accenture find, but companies may be paying for it anyway

Most of the work companies want AI agents to do could easily run on cheaper models but some are using their most expensive model for anything and everything.

Most work doesn’t need the Ferrari of AI, Wharton and Accenture find, but companies may be paying for it anyway

A new joint report from Wharton School and Accenture suggests companies may be overpaying for artificial intelligence (AI) agents by using the most powerful models for every task. The landmark study analyzed over 18,000 tasks across 923 occupations and found that more than 90% of tasks are routine enough to be handled by less expensive models.

Only 6% to 11% of work that AI agents could do is complex enough to require a cutting-edge "frontier" model, according to the findings. The report's coauthors, Eric Bradlow of Wharton and James Crowley of Accenture, emphasized that companies often do not differentiate between using the most sophisticated models for all workflow tasks and deploying different models for different parts of the workflow.

They warned that while AI compute is among the largest costs in most industries, the total spending tends to rise even as per-task cost decreases, due to the need for AI to perform increasingly diverse tasks. Bradlow and Crowley urged companies to track AI costs alongside the value it generates, suggesting the implementation of an AI P&L, or profit and loss statement, to better understand the financial impact of AI adoption.

Written by urgent.news from Fortune's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at fortune.com →

More in AI

More from Monday 16 November →