There’s an idea in machine learning called ‘the unreasonable effectiveness of data’. It describes the tendency of all models to converge on accurate representations, given enough data.
This means that data is the thing that matters to a machine trying to learn. Not the algorithms (necessarily), not the technical chops of the engineering team, and certainly not the intentions of the stakeholders.
You can infer the importance of raw data by looking at acquisitions histories - Google bought Deep Mind. The company with all the data bought the company with all the new ideas.
If you fit a model to predict which organisations would have the most advanced AI research teams, you could probably get to a reasonable accuracy using a single feature - how much data do they collect about their users?
Google and Facebook are leading the charge in AI through necessity. While other companies (like Apple, and airbnb, and Uber) are busy making products or selling tangible services, the core business models of Google and Facebook are web-based all the way down.
What you or I do with Apple’s products goes (less and less) unnoticed by Apple themselves. What we do in rented apartments is unknown to airbnb. Uber knows only the As and Bs of our journeys. Facebook and Google, on the other hand, know a whole lot more.
These two companies have become platforms for our entire lives. They’re the first places we go to get information or make a social connection. The other companies just provide us with products and services.
Google, Facebook, Apple, airbnb, Uber, Microsoft, Amazon - these companies are all filled with super-smart people, but Google and Facebook will win the AI battle.
We hear about the skills shortage and the radical salaries for research-level talent in machine learning and we think it’s all about unique ideas or algorithms. It’s not. Data is the fuel on the fire.