MICROSOFT CEO: HOW TO DEFINE A COMPANY ' S MOAT IN THE AI ERA
It's not a model. It's a closed circle.
Original title: A front without an example is not stable
Original by Satya Nadella, Microsoft CEO
Original language: Peggy
According to the editor: Microsoft CEO Satya Nadella, the real competitiveness of the AI-era business is not about which of the strongest models to place in it, but about whether it can sink its own work flow, field knowledge, organizational judgement and staff experience into an evolving learning system. In other words, enterprises cannot simply buy AI capabilities, but have their own "learning closed loops" (systems where human experience, business processes and modelling capabilities are increasingly mutually reinforcing)。
Under this framework, companies in the future will accumulate two types of capital simultaneously: human capital, i.e. the knowledge, judgement, networking, creativity and model recognition of their employees; and Token Capital (the AI capacity built and owned by the enterprise itself). Nadella emphasizes that AI does not devalue human capital, but rather makes targeting, cross-cutting connectivity and critical model identification more important. There is no one to pull, and the calculus is simply spinning in situ; there is no organization to sink its own knowledge, and models are more powerful than external tools。
THE CENTRAL JUDGEMENT OF THIS PAPER IS THAT THE FRONTIER WITHOUT ECOLOGICAL SUPPORT WILL NOT BE A STABLE FUTURE. THE VALUE OF AI SHOULD NOT BE SWALLOWED UP BY A FEW GENERIC MODELS, BUT RATHER SHOULD FORM A FRONTIER ECOSYSTEM WHERE EACH COMPANY, EACH INDUSTRY AND EVERY COUNTRY HAS ITS OWN LEARNING LOOP. BUSINESSES NEED TO BUILD A PRIVATE ASSESSMENT, AN ENHANCED PRIVATE LEARNING ENVIRONMENT AND A SEARCHABLE KNOWLEDGE BASE TO TRANSFORM HIDDEN EXPERIENCES INTO REUSABLE, SCALABLE AND ITERATIVE SYSTEMS. THE TRUE MOAT MAY NOT BE A MODEL PER SE, BUT EVEN IF IT IS REPLACED BY A GENERIC MODEL, THE ENTERPRISE WILL NOT LOSE ITS OWN "OLD COMPANY EMPLOYEE" EXPERIENCE。
THIS IS ALSO THE KEY TO CORPORATE SOVEREIGNTY IN THE AI ERA: WHO CAN TURN ORGANIZATIONAL KNOWLEDGE INTO A SYSTEM OF CONTINUOUS COMPOUNDING, WHO CAN RETAIN IP IN THE RAPID AND ITERATIVE FUTURE OF MODELS, EXPAND EMPLOYEE CAPABILITIES, AND LEAVE THE ECONOMIC VALUE OF AI IN THEIR OWN BUSINESS, INDUSTRY, AND COMMUNITY。
The following is the original text:
I'VE BEEN THINKING LATELY ABOUT WHAT THE FUTURE OF A BUSINESS WOULD BE IN AN AI-DRIVEN ECONOMY。
This transition is different from any previous migration of platforms. In the past, we have used digital systems to enhance human capital; this time, for the first time, we have been able to establish a true cognitive loop between people and digital systems. This is a very subversive thing, because it changes the way we understand "work" within companies。
THE REAL KEY QUESTION IS NOT HOW A DIGITAL TOOL OR SYSTEM IS USED, BUT HOW THE ORGANIZATION CONTINUES TO LEARN, ACCUMULATES INTELLECTUAL PROPERTY RIGHTS, DIVERSIFYS AND PROSPERS IN A WORLD WHERE THE AI MODEL CAN CONTINUOUSLY ABSORB AND COMMERCIALIZE HUMAN AND ORGANIZATIONAL EXPERTISE。
Every company must build what I call human capital and Token capital. Human capital includes staff knowledge, judgement, networking, creativity and model recognition; and Token capital is the AI capacity built and owned by the enterprise itself。
It is important that human capital does not become irrelevant as Token capital grows. On the contrary, it only becomes more important. I believe that human mobility will be at the heart of Token ' s capital growth. Human beings set ambitious targets, cross-cutting links, establish relationships and identify truly important models. Without human diversion, the calculus can only turn in place。
This means that the real opportunity lies not in choosing the best model, but in creating a learning loop on the model, where human capital and Token capital can grow back together. You can outsource a job or even a job, but you can never outsource your studies. The future of an enterprise lies in the sustainability of this learning between people and AI。
This requires a new structural approach: each enterprise should be able to build intelligent systems that will improve over time, while still retaining control over its intellectual property rights. A company should be able to replace a “generalist” model without losing its “old company employee” professional experience, which has sunk in its learning system. This will be a key test for measuring corporate control and sovereignty in the future。
Enterprises need to translate their own work streams, field knowledge and long-established judgement into AI systems that can be continuously improved in each use. Private evaluations should measure whether the model is really better at business outcomes of interest to the enterprise than just external benchmarking tests. The private intensive learning environment should make models stronger based on real tracks within the organization. The corporate knowledge base makes institutional memory searchable and token more efficient。
This loop will become the new intellectual property of the enterprise. I think it's a climber machine. Moreover, unlike most assets, it would increase in value. Each improvement in the flow of work produces better training signals, which in turn accelerates the accumulation of the unique and hidden knowledge of enterprises. Those companies that have established the system earlier will gain an advantage that is difficult to replicate, regardless of how individual model capabilities will be broken in the future。
THE LAST THING WE WANT TO SEE IS A WORLD IN WHICH EVERY COMPANY IN ALL WALKS OF LIFE GIVES VALUE TO A FEW MODELS THAT DEVOUR EVERYTHING. IF ALL VALUES WERE EVENTUALLY CAPTURED BY A FEW MODELS, THE POLITICAL-ECONOMIC STRUCTURE WOULD NOT TOLERATE SUCH AN OUTCOME AT ALL. AN AI THAT EMPTYS THE ENTIRE INDUSTRY OF THE FUTURE IS UNLIKELY TO BE LICENSED AT THE SOCIAL LEVEL。
THINK ABOUT WHAT HAPPENED IN THE FIRST PHASE OF GLOBALIZATION: THE WHOLE INDUSTRIAL ECONOMY WAS OUTSOURCED. ON THE FACE OF IT, GDP FIGURES SEEM GOOD, BUT REAL INDUSTRIAL SHIFTS AND EMPLOYMENT SHOCKS DO EXIST AND THEIR CONSEQUENCES ARE STILL BEING FELT. WE CANNOT BRING THIS DYNAMIC INTO THE AI ERA — ALLOWING A FEW AI SYSTEMS TO CAPTURE THE FULL ECONOMIC RETURNS, WHILE THE WHOLE INDUSTRY’S KNOWLEDGE IS COMMODIFIED AND EMPTIED AT THEIR FEET。
In my view, our priority must be to construct a frontier ecology, not just a frontier model. Only then can the value flow widely to every company, industry and country. In such an ecology, each organization can have its own learning loops, encode its own institutional knowledge and allow human capital to grow in combination with Token capital。
This is also the spirit of the platform that I have always shared: the value created above the platform should be greater than the value captured by the platform itself; each company should be able to innovate and create its own value。
When this is achieved, enterprises will create value for themselves and for the economic environment in which they live. The professional competence of employees is amplified, and their judgement becomes part of the system and can be replicated and scaled up, and these gains flow back to the company and its surrounding communities。
This is how firms create value for themselves and for the wider economy. It is also a stable balance that we should build together。
