A change is coming to AI.

I will try to make this article concise and to the point. As I reflect on everything I want to say about this topic that goal quickly seems unachievable, but here I go.

Data science is my passion. I have been doing it and machine learning for well over a decade across a dozen different fields from genomics, medical and insurance to signal processing and computer vision. I have taught, mentored, advised and consulted in data science, as well as interviewed over a hundred data scientists and this has lead me to two consistent yet dichotomous observations.

The first is the magic of data science/machine learning. It has the power to significantly impact every industry in the world; its potential is nearly unfathomable. The second observation is the overall disappointment with data science within companies. This seems to be in stark contrast to the first observation and it is; however, it is still true.

The reasons for the disappointment with data science and machine learning are many and not the focus of this article. Before I continue, I would like to clarify that I am making a generalization – there are a number of companies that are extremely happy with the value data science is bringing them. Now back to the point. I believe that the biggest problem with data science is a supply and demand problem; i.e., too few data scientists that are experts in the many areas needed, especially deep learning, for the large number of companies that need them. I believe cranking out more data scientists from our education system will help but not solve this problem; I have a different spin on the issue. The root cause is that machine learning is still an immature technology and as a result, the requirements to be a great data scientist are unrealistic (i.e. a unicorn, and you cannot mass produce unicorns). This complexity exaggerates the amount of time needed to complete data science projects and maximizes their risk (unknown success).

Advances in data science and machine learning will lower these requirements and they will become more realistic. Reducing complexity will solve the supply and demand problem by allowing a larger percentage of the population to be effective in these types of positions. This change is inevitable. It has been repeated over and over in history across literally every single field. New fields always begin with a high level of difficulty and thus a small number of people that excel in it. As these fields evolve and mature, the level of difficulty decreases and the number of proficient individuals naturally increases.

This change can be concerning to the very best data scientists. Today they belong to an exclusive group of experts and command large salaries. However, they need not fear, all you need to do is look back in history to see that no matter how mature a field becomes, there is always room for stars and leaders. In no way do I believe this change is bad for data science or will put data scientists out of a job. On the contrary, this is the time for data science to finally deliver on all of its promises. The final result will be happy companies which in turn will mean happy data scientists.

Part of this magical change in data science is Automated Machine Learning (AML); this will have a significant impact on reducing its complexity. First, the time needed for projects will drastically decrease. Airbnb recently confirmed this by stating, “we believe that in certain cases AML can vastly increase a data scientist’s productivity, often by an order of magnitude”. Second, the reduction in specialization and expertise needed to be coupled with the decrease in time to complete projects will minimize, and potentially eliminate their high-risk nature. Given that, it will become easier to find talent, their cost will be lower and the time invested in exploring projects dramatically reduced.

Airbnb states improvements in the quality of their models through the use of AML as well. How much better can AML be? We believe a lot.

What would it take for AI to advance far beyond the level of humans when it comes to using AI itself?

What if you could transform the growth and learnings of a data scientist, throughout his career, into an optimizable function? Imagine the possibilities…

We have.

Ople.ai.

Bringing the true power of AI to the world.

Pedro Alves, Founder and CEO.

Related Post

Inside Ople - Ted Lee

SXSW 2018 - Highlights