It would be naive to think that just our personal lives are dramatically enhanced by the recent progress in artificial intelligence. Technology is changing the way we interact with the world, and there is no question that machine learning will change how our 9-to-5 jobs look in a few years.
As a founding partner at a venture studio, we embrace these trends and try to help startups focus on the future, rather than how technology works today. The software we create will bring new opportunities and affect markets tomorrow. And the best way to prepare for change is to understand what is driving it and what trends are the most distinguished in our strive for technological singularity. I have outlined a few that I have seen recently below:
Big companies are trying to outsource their processes to machines, or at least break them apart into pieces that can be automated. This may be a major concern to an unskilled layer of workers. It can also be a stimulus for spending those saved costs on more qualified specialists in the missing areas instead.
But how do machines make certain decisions? Even Google engineers are out of touch with how their algorithms work now thanks to the system's complexity, which could prevent us from completely trusting AI. To clear up this issue, Defense Advanced Research Projects Agency (DARPA) recently funded research to make new a kind of artificial intelligence algorithms that can properly explain themselves and their conclusions, something similar to what we see in a new "Westworld" TV episode.
And what better way to know something than to ask directly? Conversation as a technology has been creeping its way into the picture for many years. It seems now we are finally reaching a state where it can become usable to some extent. This trend will set apart companies that are leveraging intelligent assistance from those that are still trying to better processes by simply hiring more humans.
The booming cloud infrastructure of corporate giants is becoming widely available for anyone to experiment with machine learning for almost nothing. Research that took people years and millions of dollars back in 2005 is practically free. And this opportunity is open to anyone, which creates a huge pool of homegrown talent, driving more and more accomplishments in the field. It also nurtures a healthy environment for companies to hire and experiment in the field of machine learning.
There has been a breakthrough in the mainstream application of deep learning, a robust artificial intelligence technique that mimics the way our brains work. Deep learning allows you to discover intimate relationships between different aspects of data, to the extent that human teachers can be eliminated from the equation. Previously, machine learning techniques were heavily dependent on human researchers guiding the algorithms through the unlabeled datasets -- but not anymore.
For example, Google's AI translation tools have created their own internal language that describes all human communications in a single feature field. This makes it super efficient at translating.
The key to building intelligent machines is giving them proper data from which to learn. Without the correct data, we are stuck with elementary machines that can only play checkers.
That's why more and more companies are becoming focused on data harvesting (especially in major work processes) to craft algorithms that can actually improve machines and functions. And most of the time this aspect heavily relies on human actions as an input, which again shows that humans won't be majorly extinct from the day to day operations -- at least for now.
To take a full advantage of this trend, consider documenting all existing processes and data it produces. This can help to accelerate data harvesting and make use of the most precious business element in the coming years.
For the last few years, our digital presence showed us what an environment where all our actions somehow influence our further actions (or actions of others through algorithms) might look like. This is one of the major trends that keeps accelerating us to a state where we hardly even have to make a decision before executing. Take the on-demand economy (Uber, Postmates, etc.) as an example. Machines already tell us where to go, when to go, how much to pay and even what to eat. Companies like Amazon have started recommending products to buy and Facebook suggests individuals to add as friends.
As a startup studio, we are seeing an exponential growth of new startups aiming to revolutionize how we build intelligent machines and communicate with them. Funding makes it possible to actually build and take their solutions to the market. And of course, investors are inclined to take part in a truly disruptive venture leading the artificial intelligence to real industry applications.
Artur Kiulian is a Partner at Colab, a venture studio that helps startups build and grow new products.