From data to artificial intelligence
An article in TechCrunch about Facebook’s advances in image recognition, which now allows it to create search systems based on the content that appears in them, leads me to think about the importance of the availability of data for the development of machine learning algorithms earning and artificial intelligence. Needless to say, Facebook’s capacity to develop systems to process and recognize images is about accessing tens of millions of images that have been tagged and commented on by its users those of Instagram.
Thinking about the possibilities of artificial intelligence for our business, we need to start with the data we’re going to analyze, but not all data is created equal: we need formats and tools that are sufficiently open to allow their processing, something that is not always easy for companies that, for a long time, processed their data in legacy systems that are not easy to integrate.
Coming from a time in which many industries have been concerned about catching up on issues related to big data facilitates this task to some extent: when you already have data scientists on staff, the least you can expect is that they have carried out the cleaning and cataloging of the data sources they intend to include in their analytics. But after big data comes the next step: artificial intelligence, advances in the field of which are leading data scientists to realize that they need to evolve into that discipline or find themselves obsolete.
Data is the gasoline that powers artificial intelligence. Data allows us to develop the best algorithms, and above all, to improve them over time so that they produce better results and adapt to changing conditions. The availability of more and more data in autonomous driving as fleets of test vehicles run up more and more kilometers is what allows Tesla to reduce the number of disengagements, episodes in which the driver is forced to take control, to the current levels: between October and November of last 2016, four autonomous vehicles of the company covered 885 km on Californian roads, experiencing 182 of those moments, allowing the company to continue improving on the basis of that accumulated experience. In contrast, Waymo, which has all the data accumulated by Google’s experience with autonomous driving and always conceived their project as fully autonomous, managed throughout 2016 to reduce the number of these disengagements from 0.8 per thousand miles to 0.2: an impressive progress fueled, again, by the availability of data from the very beginning of the project.
The biggest mistake that can be made in artificial intelligence is to try to judge an algorithm by its results the moment we get it, without taking into account the progress that can be made by using more and better data. Criticizing Amazon’s Echo by saying that it is little more than a glorified radio clock overlooks the fundamental thing: that with eight million devices on the market, Amazon’s chances of improving Echo’s intelligence are practically unlimited, and that means that over time we will understand things better, gradually reducing its errors, meaning that it will soon be a device that we will end up wondering how we ever lived without it.
If there is any sport where artificial intelligence could be used as a referee, the immediate candidate becomes American football, where everything is quantified, analyzed and processed to the nth degree.
Similarly, insurance companies will benefit from the savings and enhancements of artificial intelligence-based expertise as long as they have large amounts of data correctly stored and structured.
And of course academic institutions will be able to use artificial intelligence in the educational process, providing they have complete files, correctly structured and prepared so as to be treated. And I can assure you that from my experience, this is not the case with most business schools and universities.
Understanding the evolution of data toward machine learning and artificial intelligence is now increasingly important for decision makers, and increasingly strategic for businesses.
This is how some companies will end up on one side of the new digital divide or the other.
(En español, aquí)
