The "Egg Theory" of AI Agents

🌈 Abstract

The article discusses the intersection of technology and human psychology, particularly in the context of AI agents and their integration into consumer and enterprise workflows. It explores the "egg theory" and the IKEA effect, which suggest that people value things more when they feel they have contributed to the process. The article provides examples of how AI agents can be designed to maintain a sense of human involvement and control, rather than completely removing the human from the loop.

🙋 Q&A

[01] The "Egg Theory" and the IKEA Effect

1. What is the "egg theory" in consumer psychology, and how does it relate to the adoption of AI agents? The "egg theory" suggests that when instant cake mixes were first introduced, they sold poorly because people felt guilty about not contributing to the baking process. By requiring the addition of an egg, people felt like they had contributed, and sales soared. This theory can be applied to the adoption of AI agents - people may feel uncomfortable with AI agents that completely remove them from the workflow, as they want to feel like they are contributing or maintaining a sense of control.

2. What is the IKEA effect, and how does it relate to the adoption of AI agents? The IKEA effect is a cognitive bias that helps explain why people place higher value on things they helped to build or create. This relates to the adoption of AI agents, as people may value products more if they feel they have some level of involvement or control in the process, rather than being completely removed from the workflow.

[02] Integrating AI Agents into Consumer and Enterprise Workflows

1. How can AI agents be designed to maintain a sense of human involvement and control in a consumer workflow, such as booking a trip? The article suggests that a better approach would be for the AI agent to present the user with a few options to choose from, rather than completely automating the process. This allows the user to feel in-the-loop and in control, while the agent handles the grunt work.

2. How can AI agents be integrated into enterprise workflows while maintaining a sense of human involvement and control? In the enterprise context, the article suggests that the best interface would include approvals by the human user before payments are released. This gives the user comfort and a sense of involvement in an important part of their business, even though the AI agent may be better equipped to handle the task.

3. What is the difference between an "AI agent" and an "AI copilot", and why might the latter be more palatable to people? The article suggests that the term "AI copilot" implies more of a human augmentation rather than human obfuscation, which may be more palatable to people as they adjust to the reality of software doing work for them.

[03] Balancing Technology and Human Involvement

1. Why is it important for companies to embed human decision-making into workflows, rather than removing the need for human input altogether? The article suggests that removing the human from the workflow entirely can produce a worse user experience, as people may feel uncomfortable with too much change too fast. Injecting a bit of friction and maintaining moments of human involvement can help keep users happy and improve retention.

2. What are some examples of how companies can balance technology and human involvement to create winning products? The article provides examples of how Midjourney allows users to contribute to the creative process by writing prompts and selecting from generated designs, and how AI agents in travel booking can present options for the user to choose from, rather than making the decision entirely on their own.