Artificial intelligence and the future of companies:.
10 self-evident conclusions.
OpenAI's Hemant Asirvatam and Panni Moksky are leading 24-year-old teams of agents whose management has been brought down to their level. Their essay focuses on the transition period in which artificial intelligence integrates into ordinary companies in three ways - under people, alongside them and between levels of management.
challenges.
1. The stairs may collapse from above. The work of a manager - reading, meetings, analysis, decisions on the computer - and models may already be closer to it than operations work. Some of the management is automated before the work is managed. There are almost no tests for executive coordination, so no one knows how close it is.
2. AI pulls decisions to the top. In the past, information flowed slowly and decisions had to be relayed. An internal AI consultant shows the manager the whole company in real time, and the reason for delegation is weakened.
3. Salaries may diverge further. An ordinary worker with a team of agents is responsible for the amount of work that the head of the department used to have, and his good or bad choice weighs just as much. Those who manage their agents better will get more. What sets the gap apart is that the AI gives equally good advice to everyone.
what will work
4. Whoever did it himself is a representative. While AI can't show you what you don't know, it will benefit a performer who knows the method. A leader without this knowledge will not discover even a strong model.
5. Artificial intelligence consultant to the manager. In order to understand what was actually happening in the company, the manager called in external consultants - expensively and for several months. The model reads correspondence, documents and code of all departments and does it all the time and almost free of charge. It does not require technical knowledge, so the implementation of AI can start in the executive office.
6. Aim less, shoot more often. Approvals and roadmaps were based on the fact that it is more expensive to do than to ask. Prototyping becomes cheaper, and some of the appropriate levels of middle management become redundant.
7.