Architects and engineers among professions most automatable by AI according to Anthropic

A study by AI company Anthropic has found that much work done by architects and engineers could be done twice as quickly with large language models, but claims there has been little impact on jobs so far.

Released last week, the paper details the differences between actual and theoretical task replacement by artificial intelligence (AI) programmes, based on a metric called "exposure".

Anthropic's study is called Labor Market Impacts of AI: A New Measure and Early Evidence. It utilised Anthropic user data from its Claude large language model (LLM) to plot observed exposure versus theoretical exposure.

Exposure for architecture among the highest

Theoretical exposure measures how much of an industry's tasks could be done twice as quickly by AI.

According to this predictive model, as much as 70 per cent of the architecture and engineering fields could be affected. The arts, design and media sector also has a high level of exposure, with 68 per cent of the sector potentially affected.

The data for theoretical exposure were derived from a study conducted by Tyna Eloundou, Sam Manning, Pamela Mishkin, and Daniel Rock. Eloundou and Mishkin are employees of another major AI firm, OpenAI.

Though the theoretical exposure for architecture and engineering was among the highest, the study showed that the observed exposure was quite low, close to zero.

"High exposure has not yet correlated with unemployment"

This aligns with an AIA study released last year that found only six per cent of practicing architects were regularly using AI.

"The coverage shows AI is far from reaching its theoretical capabilities," said the Anthropic study. "High exposure has not yet correlated with unemployment."

The only group that may have been affected, according to the study, are people between the ages of 22-25 entering the job market, though the study could not correlate this with the effects of AI.

"Using survey data from the US, we find no impact on unemployment rates for workers in the most exposed occupations, although there’s tentative evidence that hiring into those professions has slowed slightly for workers aged 22-25," said the study, which referenced US labor stastitics.

However, the model may well serve for predicting future unemployment as a result of AI job exposure.

Since the release of early consumer LLM programmes in 2022, the architecture field has been asking what effects automation by AI might have on the profession.

Earlier this year, the first home designed by a humanoid designer was showcased in Denmark, while last year Tim Fu released renderings of a masterplan he claimed used AI in all its stages.

The post Architects and engineers among professions most automatable by AI according to Anthropic appeared first on Dezeen.