Democratization and Interpretability: DataRobot’s 2019 AI Predictions

DataRobot CEO and Co-Founder Jeremy Achin recently spoke with Information Ageabout the future of automation for 2019. (Read the full interview here.) Jeremy states that even if everyone on the planet became a data scientist, there still wouldn't be enough resources to solve every business problem. Automation is the key to helping businesses and data scientists, and 2019 will be a big year for automation.

In the interview, Jeremy focuses on two predictions for 2019:

Firstly, thanks to automation, data scientists will produce more, a lot more.

Second, he sees the democratization of the role [data scientists]. With automation, people in other areas, what Jeremy calls AI workers, software engineers, data engineers and business analysts, can work more closely with data scientists.

Additionally, we spoke with other DataRobot employees about their predictions for AI and machine learning in 2019. Two major themes became apparent: the democratization of data science and interpretability of AI models.

Through the democratization of AI, data science responsibilities within organizations are becoming more accessible to other roles, such as business analysts. Automation tools now enable others to build models and make predictions of their own. Additionally. these tools  empower data scientists to focus on iteration and innovation - helping them become more productive.

Tristan Spaulding, Director of Product Management, predicts that in 2019 we will see the continued rise of business analysts adopting machine learning, exceeding the number of data scientists.

“For the first time, the number of business analytics professionals applying machine learning models will exceed the number of data scientists doing so.

New adopters of machine learning will painfully discover that building a model is only the first leg of the journey—deploying and maintaining models takes just as much time and effort.”

— Tristan Spaulding

 

Interpretability and trust are two important themes for 2019, as well as something our team strives to promote and strengthen. Peter Simon, a DataRobot Data Scientist, shares his thoughts about the importance of interpretability in 2019.

“2019 will be the year in which interpretability of AI will be paramount - not necessarily in terms of understanding the underlying mechanisms of how AI works, but certainly in terms of how an AI system's outputs are driven by its inputs. With the recent focus on bias in AI and its link to the underlying human behaviors being modelled, technologies which cannot provide clear explanations of their outputs will be left behind.”

— Peter Simon

 

Rajiv Shah, a DataRobot Data Scientist, makes a prediction about the value and importance of regulation around AI, as well as ensuring accountability and accessibility by both AI regulators and users.

“2019 will be a year with an increased call for regulation of AI, especially as it comes to explaining the decisions of models.

You will see an increased focus on accountability around AI, whether within the industry or by governments. One tangible outcome will be a push for more interpretable and explainable models. This will become evident in the wider use of explainability techniques by data scientists, going all the way to regulators encouraging or requiring a wider set of AI be easily understood by both regulators and users.”

— Rajiv Shah, Data Scientist

 

For more information, visit: https://blog.datarobot.com/democratization-and-interpretability-datarobots-2019-ai-predictions?utm_medium=(not%20set)&utm_source=google.com&utm_campaign=(not%20set)&utm_term=(not%20provided)