Augment not replace workers

Listened to a great presentation on The Impact of Generative AI on Workforce Productivity. It made a compelling case for using AI to augment human workers rather than replace them.

Brynjolfsson and McAfee’s recent research indicates that around 80% of the US workforce will have more than 10% of their work impacted by generative AI technologies like large language models. These systems like GPT-3, DALL-E, and others have demonstrated the ability to generate human-like text, images, code, and more.

Through their company Workhelix, Brynjolfsson and McAfee have been researching how different industries, job functions, and specific work tasks are likely to be affected as generative AI rolls out more broadly. They are advising multiple Fortune 500 firms on strategies for leveraging AI to boost productivity rather than reduce headcount.

In this webinar presentation, they provided insightful forecasts for how generative AI will spread across the economy in the coming years. They highlighted which functions like customer service and document creation are likely to see the earliest and most dramatic impacts. The presenters also called out industries like law, healthcare, and education that will need to thoughtfully integrate AI into their workflows.

A core argument Brynjolfsson and McAfee make is that instead of automating jobs away, companies should focus on using AI to augment human strengths. With the right strategies, workers can become more productive, creative, and valuable with the help of AI. This includes using AI for tasks like information retrieval and data processing while preserving roles that require emotional intelligence, strategy, and advanced expertise.

The presentation offers a cautiously optimistic perspective on the future of work in the AI age. While disruption is inevitable, Brynjolfsson and McAfee emphasize that we have the agency to shape how these technologies impact jobs and identify new opportunities for humans to thrive with AI as a collaborative tool. Companies that embrace this augmented workforce approach are poised to gain competitive advantage.

Can augmenting workers be made to work?

While augmentation has benefits, its feasibility and ultimate employment impacts remain uncertain. Reasonable counterarguments exist that greater workforce disruption may occur regardless of augmentation strategies.

• Automation pessimism - Some argue that AI will inevitably displace large segments of the workforce and lead to widespread technological unemployment, regardless of augmentation strategies. The pace of advancement in generative AI makes more jobs automatable.
• Implementation challenges - While augmentation sounds good in theory, actually integrating AI smoothly into human workflows poses major practical challenges. Many tasks are too interdependent for clean augmentation.
• Cost incentives - Replacing human roles with AI is seen by some companies as a way to reduce costs, not just boost productivity. The drive to cut expenses through automation overrides augmentation goals.
• Structural inertia - Existing corporate structures, job roles, training programs, employee policies etc. may be too rigid to adapt to an augmentation model. Significant restructuring would be needed.
• Labor concerns - Augmentation could place more monitoring and performance management pressures on workers. Some may oppose AI encroaching further into human roles and responsibilities.
• Overreliance risks - Becoming over-dependent on AI augmentation could deskill workers, make companies more vulnerable to AI failures/errors, and exacerbate layoffs during downturns.
• Short-term bias - Shareholder demands for near-term profitability may override longer-term workforce planning and employee retention/retraining efforts needed for augmentation.
• Task generalizability - Not all tasks can be cleanly decomposed into an “AI part” and “human part” needed for effective augmentation models. The interplay is more complex.