Follow Datanami:
January 6, 2020

2020 AI Trends for the Enterprise

In looking back at how far we’ve come in the last decade, from BI to predictive analytics to AI, one thing is for sure: 2020 will undoubtedly hold even more rapid development in the field of data science and machine learning. Aside from generally preparing AI strategies for near constant change (i.e., introducing flexibility and sustainability across technology,processes, and even people), what can enterprises expect in this new decade?

A Change in the Data Scientist Role Itself

All data scientists will almost certainly see a change in their roles in the coming decade in large part because of shifts in organizational approaches to data science, including a focus on data democratization, as well as technological developments, like investment in AI systems to facilitate AutoML and leverage automation for larger swaths of the model development process.

That means from a business perspective, organizations will have to be prepared to look for specialists. Indeed, data science generalists will become less desirable because companies will be looking for specific skill sets that fill defined holes in their AI deployment strategy (like NLP, model monitoring, operationalization, etc.).

The Essential Management of Cloud Costs

There’s no question that across all industries, a great migration to the cloud is underway for businesses worldwide. Dataiku conducted a survey in mid-2019 among IT professionals in large enterprises across industries and found that just less than a quarter of them are storing data used for machine learning projects exclusively on premise.

In order to keep cloud costs under control in 2020 and beyond, organizations will have to start developing strategies for optimizing cloud use, including considering multi- vs. single-cloud approaches, adding a layer of transparency around the use of cloud resources, and using AI to automatically monitor, scale, and control IT infrastructure (otherwise known as AIOps).

The Move Toward Initiative-Driven Teams

In 2019, the data science center of excellence took center stage. Companies around the world and across industries created centralized teams to handle data initiatives as a way to kick-start AI efforts (for just one example, see Nicholas Bignell, Director of Data Science at UBS, talk about spinning up their center of excellence).

The trend for 2020 will be moving toward a more formalized way for data experts and business experts to work together: the initiative-driven teams model. Along the theme of elasticity, this approach allows teams to be spun up or spun down based on the project for tight, expert focus, ensuring that the results align well with business needs and expectations.

Data experts still work on a variety of projects, so the creativity problem from the embedded model is not an issue. And once the project is finished, there is a large data team that can handle maintenance and monitoring of the project instead of that falling back on the business (which is not their area of expertise and might prevent them from scaling out the number of projects they deploy).

Want to know what else is on tap for 2020 and beyond? For a technical perspective, check out what Dataiku data scientists have to say. For an organizational perspective, don’t miss the live webinar on January 8 (available after on demand).

 

Datanami