The guilt of AI productivity

“We have exciting times ahead. My productivity has increased nearly 10X compared to the pre-LLM era. It has been so much less stressful doing the everyday work I used to sweat over. I seem to be more in control of my tasks. I don’t know how I managed before Generative AI came into my life.”

“Who all have exciting times ahead? By whose yardstick? How many of them are there like you? Aren’t you talking about only a few? Ask how most of the population is living without leveraging Generative AI’s benefits.”

I felt selfish. This is something we must discuss this week. Should we discuss AI? Or my selfishness? Both. There are multiple ways in which the benefits of the AI era flow.

First, the product or service provider does more in the same amount of time. They sell more products and services. If the provider earns per product, service, value-added, or outcome rather than on effort spent, they have a lot to gain. Second, customers are watching. Why would they agree to pay for something that was created “easily”, expending less effort? They would like a share of the advantages. Third, imagine nothing changes for the provider and the customer. The developer continues to deliver at the earlier pace but is building faster. They could utilise the saved time for upskilling, reading, doing things they always wanted to do but never could, and maintaining a better work-life balance. This would happen when neither the employer nor the customer is fully aware of the benefits of AI.

Fourth, the same amount of work gets done with fewer workers. The pace of delivery remains the same as earlier. AI-aware customers would demand a reduction in product and service prices. The employer spends less. However,

by doing this, you have just ensured that workers’ stress and difficulty levels remain unchanged. The per-worker share of work increases to compensate for the laid-off workers.

Unfortunately, we see the fourth scenario trending in the industry these days. In my opinion, companies that adopt this approach to reaping AI benefits risk their long-term growth. They might be focusing on only a short-term reaction to the rapid pace of developments.

Fifth, there is a divide between the AI haves and have-nots. One organisation cannot assume the same level of productivity increase from its entire workforce. A realistic distribution of AI literacy must be considered. Sixth, expectations from customers and employers increase, often creating a disequilibrium with the actual worker’s capability. Seventh, the escalation in output quality might lead to the creation of new markets.

Eighth, and my most favourite one, is the rebounding focus on human premium. When AI-generated content floods online writing platforms, the training datasets we use to build the next large language model (LLM) will also be AI-generated. The next time content is created, the output will be AI-AI-generated. Where do we stop the chain? With AI-AI-AI-AI-generated content? Hence, I must keep honing my human content creation skills until the rebound happens and I start commanding a premium for my work. Again.

Ultimately, the AI era is less about raw productivity gains and more about how equitably and thoughtfully those gains are distributed. If we optimise only for efficiency, we risk widening gaps between firms and workers, between AI-native and AI-excluded, and between output and meaning. The real opportunity lies in redesigning incentives to reward learning, augment human capability, and preserve the distinct value of human judgment and creativity. Generative AI should not merely compress effort into output but expand what individuals and organisations can aspire to do. The question is not whether AI benefits us, but who benefits, how much, and at what long-term cost.

END OF ARTICLE