What Meta’s Cicero and bots that code could mean for highly skilled workers

Article by Jacob Bourne | Nov 28, 2022

The data: Automation is in part to blame for worsening wage inequality in the US over the past 40 years, according to MIT research.

US men without high school degrees earned 15% less in 2016 than in 1980, accounting for inflation, despite the US GDP growing from $6.82 trillion in 1980 to $18.7 trillion in 2016, per New Atlas.
Rapid automation of routine tasks in industries is responsible for 50% to 70% of the increase in US wage disparities over the last four decades, the MIT economists concluded.

What it means for the elite workforce: Using AI, Big Data, robotics, cloud computing, and rising compute power to fulfill more workplace roles could chip away at tasks done by the most educated, highly skilled workers.

Three recent examples illustrate the threat:

Meta AI’s Cicero program achieved more than double the average score of human opponents in the strategic board game Diplomacy using skills like natural language, negotiation, persuasion, and functional “empathy.” AI’s ability to outwit humans in a game of social strategizing, coupled with data analytics, could enable it to perform tasks currently fulfilled by business executives and government officials.

Google has demonstrated that it can use large language models (LLMs) to teach robots how to write their own code as part of its PaLM-SayCan research, which could endow them with complex reasoning, technical, and physical skills useful in a variety of industries.
And MIT researchers building self-assembling robots—alongside advances in end-of-arm tooling, sensors, and computer vision—could soon eliminate the need for human dexterity in advanced manufacturing.

Automation management is the future of work: The tight labor market is fueling the push for automation as corporations struggle to meet profit margin expectations amid growth headwinds.

The increasing prowess of AI and robotics makes the future of work a focal point.
This year’s debut of generative AI that can create art, music, write, code, and more has debunked robotics industry leaders’ refrain that automation will free humans to pursue creative work.
But a myriad of AI programs—like Meta’s Galactica AI—that presented false and biased information as scientific fact gives a glimpse of people’s role in the fourth industrial revolution.
Lacking human values, advanced AI-powered programs and machines will require human management and surveillance, much of which will likely take place in the cloud and could result in an influx of new jobs.
This oversight role may be challenged by AIs learning to cheat, as Oxford University and Australian National University researchers have warned. Meta’s Cicero gives credence to the possibility.