The State of AI Development and Operations in 2019
October 01, 2019

Mark Coleman
Dotscience

Share this

The use of AI is booming across the modern enterprise. In fact, according to Gartner's 2019 CIO Survey, the number of enterprises implementing AI grew 270% in the past four years and tripled in the past year. However, many enterprises will be unable to realize the full potential of their initiatives until they find more efficient means of tracking data, code, models and metrics across the entire AI lifecycle.

To better understand the AI maturity of businesses, Dotscience conducted a survey of 500 industry professionals in its inaugural State of Development and Operations of AI Applications 2019 report.

Research findings indicate that although enterprises are dedicating significant time and resources towards their AI deployments, many data science and ML teams don't have the adequate tools needed to properly collaborate on, build and deploy AI models efficiently.

AI Goes Mainstream

AI has moved beyond the experimentation stage and is now seen as a critical and impactful function for many businesses. Enterprises are becoming increasingly reliant on AI for its ability to deliver greater operational efficiency, streamline complex business processes, and support cost control and profit potential. This is evidenced by the survey results, which indicate that the top three drivers of AI adoption are efficiency gains (47%), growth initiatives (46%) and digital transformation (44%). Furthermore, over 88% of respondents at organizations where AI is in production indicated that AI has either been impactful or highly impactful to their company's competitive advantage. The exponential growth of AI's value and influence is also reflected in the large investments organizations are making in AI. Nearly a third of respondents (30%) are budgeting between 1 and 10 million dollars for AI tools, platforms and services.

Unfortunately, it's not all rainbows and sunshine in the world of enterprise AI. The study also found that despite this level of financial commitment, data science and ML teams continue to experience issues, including duplicating their work (33%), rewriting models after team members leave (28%), justifying the value of their projects to the wider business (27%), and slow and unpredictable AI projects (25%).

Manual Tools and Processes

Despite providing an impactful competitive advantage for enterprises, AI deployments today are largely slow and inefficient. The manual tools and processes primarily in use to operationalize ML and AI don't support the scaling and governance demanded of many AI initiatives.

The top two ways that ML engineers and data scientists collaborate with each other are by using a manually updated shared spreadsheet for metrics (44%) and sitting in the same office and working closely together (38%). These methods of collaboration ultimately disrupt efficiency and limit AI's potential. Machine learning has many moving parts, and teams require version control for their training and test data, their code and their environment, as well as metrics and hyperparameters in order to collaborate efficiently. Survey findings show that over 35% of organizations don't use any version control for their training and test data. However, of those who don't currently have any version control, over 60% would like to.

These limitations are compounded by the fact that nearly 90% of respondents either manually track model provenance (a complete record of all the steps taken to create an AI model) or do not track provenance at all. And of those that manually track model provenance, more than half (52%) do their tracking in a spreadsheet or wiki, a cumbersome and error-prone approach.

Challenges in Scaling AI Initiatives

Despite significant investment in AI, many companies are still struggling to stabilize and scale their AI initiatives. The manual tools and processes being used by many for AI model development are insufficient and do not support the required scaling and governance.

While 63% of businesses reported they are spending between $500,000 and $10 million on their AI efforts, 61% of respondents continue to experience a variety of operational challenges. This is evidenced by the fact that 64% of organizations deploying AI said that it is taking between 7 to 18 months to get their AI workloads from idea into production, illustrating the slow, unpredictable nature of AI projects today. Meanwhile, for nearly another 20%, the anticipated timeline is 19+ months to production.

DevOps Like It's 1999

The challenges faced by data science and ML teams today are reminiscent of the same challenges facing software engineers in the late 1990s. Then came DevOps, which transformed the way software engineers deliver applications by making it possible to collaborate, test and deliver software continuously.

With ML and AI projects today, collaboration is even more challenging when compared to basic software engineering. Normal software development tools focus on versions or commits of code whereas ML has many more moving parts. ML teams require version control for both training and test data, their code and their environment, as well as metrics and hyperparameters for each training run.

While ML and AI are understood as powerful technologies with the potential to reinvent the global economy, operationalizing AI still remains a major hurdle for many organizations. To simplify, accelerate and control every stage of the AI model lifecycle, the same DevOps-like principles of collaboration, fast feedback and continuous delivery should be applied to AI. Only then can enterprises realize the full potential of their AI deployments across the organization.

Mark Coleman is VP of Product and Marketing at Dotscience
Share this

The Latest

October 17, 2019

As the data generated by organizations grows, APM tools are now required to do a lot more than basic monitoring of metrics. Modern data is often raw and unstructured and requires more advanced methods of analysis. The tools must help dig deep into this data for both forensic analysis and predictive analysis. To extract more accurate and cheaper insights, modern APM tools use Big Data techniques to store, access, and analyze the multi-dimensional data ...

October 16, 2019

Modern enterprises are generating data at an unprecedented rate but aren't taking advantage of all the data available to them in order to drive real-time, actionable insights. According to a recent study commissioned by Actian, more than half of enterprises today are unable to efficiently manage nor effectively use data to drive decision-making ...

October 15, 2019

According to a study by Forrester Research, an enhanced UX design can increase the conversion rate by 400%. If UX has become the ultimate arbiter in determining the success or failure of a product or service, let us first understand what UX is all about ...

October 10, 2019

The requirements of an APM tool are now much more complex than they've ever been. Not only do they need to trace a user transaction across numerous microservices on the same system, but they also need to happen pretty fast ...

October 09, 2019

Performance monitoring is an old problem. As technology has advanced, we've had to evolve how we monitor applications. Initially, performance monitoring largely involved sending ICMP messages to start troubleshooting a down or slow application. Applications have gotten much more complex, so this is no longer enough. Now we need to know not just whether an application is broken, but why it broke. So APM has had to evolve over the years for us to get there. But how did this evolution take place, and what happens next? Let's find out ...

October 08, 2019

There are some IT organizations that are using DevOps methodology but are wary of getting bogged down in ITSM procedures. But without at least some ITSM controls in place, organizations lose their focus on systematic customer engagement, making it harder for them to scale ...

October 07, 2019
OK, I admit it. "Service modeling" is an awkward term, especially when you're trying to frame three rather controversial acronyms in the same overall place: CMDB, CMS and DDM. Nevertheless, that's exactly what we did in EMA's most recent research: <span style="font-style: italic;">Service Modeling in the Age of Cloud and Containers</span>. The goal was to establish a more holistic context for looking at the synergies and differences across all these areas ...
October 03, 2019

If you have deployed a Java application in production, you've probably encountered a situation where the application suddenly starts to take up a large amount of CPU. When this happens, application response becomes sluggish and users begin to complain about slow response. Often the solution to this problem is to restart the application and, lo and behold, the problem goes away — only to reappear a few days later. A key question then is: how to troubleshoot high CPU usage of a Java application? ...

October 02, 2019

Operations are no longer tethered tightly to a main office, as the headquarters-centric model has been retired in favor of a more decentralized enterprise structure. Rather than focus the business around a single location, enterprises are now comprised of a web of remote offices and individuals, where network connectivity has broken down the geographic barriers that in the past limited the availability of talent and resources. Key to the success of the decentralized enterprise model is a new generation of collaboration and communication tools ...

October 01, 2019

To better understand the AI maturity of businesses, Dotscience conducted a survey of 500 industry professionals. Research findings indicate that although enterprises are dedicating significant time and resources towards their AI deployments, many data science and ML teams don't have the adequate tools needed to properly collaborate on, build and deploy AI models efficiently ...