AI Chatbot Deployment Decision
I think they should deploy the chatbot. While it is clear that there are safety risks such as “hallucinations, data privacy, outages” mentioned by John, I agree that competitors are likely releasing similar products, and as long as appropriate safety measures are taken, they should deploy the chatbot before they are outrun by competitors. If not, there will be endless deliberation because with new technology such as AI, there is always uncertainty about its future. I think the concern that the salesbots will perform worse than the human team is not as important as safety concerns because, as mentioned in the article generative AI will learn and improve iteratively and will likely outperform humans in the future. If other companies start using salesbots first, they will gain an advantage over Jeannie’s company. AI is more likely to give satisfactory service in competence, consistency, and convenience, particularly in roles that follow scripts such as sales. I would be more hesitant if this AI was replacing more complex and nuanced tasks, but for the purposes of sales I think AI can adequately handle everyday jobs.
Decision Considerations
They should consider safety measures and rollback plans if issues arise. For example, how will they stop services if the salesbots make mistakes? Given the potential safety consequences, I think they should adopt a slow rollout solution. For example, as Dharmesh suggested, they can deploy a semi-autonomous bot that keeps humans in the loop. They can also deploy the bot to a partial, small customer base first and test it before doing a full rollout. If given the green light from customer feedback, product, engineers, and safety engineers, then I would deploy the AI salesbot fully to keep ahead the transformative AI era.
