It’s that time of the year: about a month early, Shivon Zilis from Bloomberg Beta has just released her annual State of Machine Intelligence report (“Machine Intelligence” is Shivon’s preferred term to describe Machine Learning – we’ll go with it). This time it is published in the Harvard Business Review.
We at Import.io are very proud to be listed as part of the competitive landscape for Machine Intelligence! As you would expect we appear in the Data Capture and Enrichment category.
Compared to when Shivon last published her 2015 report, there has been a massive shift in the number and make up of companies interested in Machine Intelligence. A year ago executives at Fortune 500 companies were only just starting to learn about Machine Intelligence; this year those same nervous executives have overtaken academics and startup founders as Shivon’s target audience for her report and so it appears in the Harvard Business Review this year.
Business leaders today realize that Machine Intelligence is starting to affect all industries, and that it is here to stay. (And they would rather avoid the fate of the “experts” who in the mid 90’s thought that the Internet was the new CB radio: a fad that would go away and therefore could be safely ignored – a rare opinion nowadays). I also suspect that today’s business leaders also grasp that the transformation will happen faster and be even more profound than the Internet, because it will affect both the physical and virtual worlds: we are not just adding a layer of communication on top of the physical world, we are populating it with intelligent gadgets, autonomous vehicles and drones.
3 pieces of advice and a word of caution
So what’s a business leader to do with this Machine Intelligence thing? Shivon and James have three pieces of advice and a word of caution.
Use new Machine Intelligence products made by others
You should use new tools, enabled or improved by Machine Intelligence, to make your team more productive. This may be an easy way for a traditional business to test the Machine Intelligence waters, but it is fairly limited in the benefits that it will bring and offers no significant competitive advantage.
Look to new sources of data
You can, and should, look at using new sources of data. Machine Intelligence can take advantage of just about any data: text, voice, images, videos; so you should make sure that you are looking to these sources of data when considering your corporate data strategy. With these new sources of data fed into Machine Intelligence models you can completely reinvent your products.
Understand how your software products can change
Much more importantly, Machine Intelligence is changing the nature of software. Most software can be seen as code acting on data through a set of hand-coded business rules. It is now evolving into code acting on data through a model learned from other data. It is hard to understate this key point: you no longer write rules, you grow a model. With the right tools (which are freely available and very powerful) and the right team (not so easy to get, can you say “war for talent”?) you can create models and get a leg up on the competition. Then, if you are careful to improve a model by feeding it fresh data (usually provided by your own users), the model will improve and stay up-to-date.
Exercise caution when deciding where to apply Machine Intelligence
Lastly a word of caution: you can’t assume that moving your product to use Machine Intelligence will automatically equate with success. Like human intelligence, Machine Intelligence makes mistakes and depending on the domain, you have to do a careful risk / benefit analysis when deciding whether to use Machine Intelligence. Accidentally showing the wrong ad clearly does not represent the same risk to your business as a misguided patient diagnosis or the wrong move in an autonomous vehicle.
Don’t miss the boat!
The report’s main admonition: don’t miss that boat! Those who understand what is at stake and implement the right changes first will “outrun and outlast their competitors”. We couldn’t agree more.
You should go and read the complete report for yourself. It is short but packed with information.
Turn the web into data for free
Create your own datasets in minutes, no coding required
Powerful data extraction platform
Point and click interface
Export your data in any format
Unlimited queries and APIs