Big Data

Kayak chief scientist says AI benefits outweigh deployment pain

Join executive leaders at the Data, Analytics, & Intelligent Automation Summit, presented by Accenture. Watch now!


Organizations are using AI as a tool for generating value, and they plan to invest even more in AI in response to the pandemic and its acceleration of all things digital, according to McKinsey. But work remains to scale the impact. While the percentage of firms investing greater than $50 million in AI initiatives was up to 64.8% in 2020 (from just 39.7% in 2018), only 14.6% of organizations report that they deployed AI capabilities into widespread production, a NewVantage Partners survey found.

Implementing AI can indeed be rife with challenges. But as Kayak chief scientist Matthias Keller said during a session at VentureBeat Transform 2021, successful adoption hinges on buy-in from the relevant stakeholders in an organization.

“[Organizations need to find] the most meaningful areas where they should apply AI and machine learning [to their] products. When you look at machine learning product development, there’s a higher risk, and so it takes extra consideration — understanding which are the right moves, because these projects can become very complicated,” Keller said. “We [at Kayak] are looking to invest in the areas where we need the customization and not try to reinvent the wheel for problems that have been solved … There’s a lot of value for us to go deep in developing new models or customizing existing work to make it fit for our unique use cases.”

Overcoming challenges

O’Reilly’s 2021 AI Adoption in the Enterprise report, which surveyed more than 3,500 business leaders, suggested that a significant barrier to AI adoption is a lack of quality data, with 18% of respondents saying their organization is only beginning to realize the importance of high-quality data. Participants in Alation’s State of the Data Culture Report said the same, with a clear majority of employees (87%) pegging data quality issues as the reason their organizations failed to successfully implement AI.

Kayak once faced this challenge with its internal AI development stack, according to Keller. Previously, the company was limited when it came to handling complex AI systems in real time.

“If you look at machine learning problems, you can classify them around a couple of dimensions. Not only was it extremely difficult for us to find the data needed to train our system[s], but we weren’t set up from an engineering perspective, where we could have data scientists use their tools and engineers use their tools,” Keller said. “We also weren’t able to easily retrain [machine learning models] because it involves so many people and getting the data.”

Overcoming these blockers required better understanding the needs of specific applications, as well as the engineering shortcomings that existed with the old system. But once this occurred, Kayak’s team was able to significantly streamline the way machine learning projects are ideated and implemented across the organization.

“We’re now able to have our data in days, if not even hours, and we have much more data to look at,” Keller said. “It’s billions of samples of much cleaner data. And we can now create models that have tens of millions of parameters and deploy these into production without having to write it twice. We can also automate it using continuous integration and continuous deployment and have these models being updated in whatever the right cadence is, which for us right now is somewhere in between weekly and daily.”

In concrete terms, Kayak’s new infrastructure allows it to solve more meaningful problems like searches across hundreds of travel websites at once. Before, the company’s data scientists found it impossible to send large volumes of data goes through the necessary machine learning systems and classifiers.

“[This new setup] allows us to become so much better in getting the most relevant content to the user — and also in getting that content served up to the user,” Keller said.

The business value of AI is perhaps why IBM’s Global AI Adoption Index found that enterprises plan “significant investments” in AI throughout the coming year. Adoption is being driven by both pressures and opportunities, from the pandemic to technological advances that make AI more accessible. Of the categories of AI organizations are adopting, natural language processing (NLP) is at the forefront. Almost half of businesses say they’re using apps powered by NLP and 1 in 4 organizations plans to use the technology over the course of 2021. Customer service is the top use case, with 52% of companies deploying or considering deploying NLP for it, according to IBM.

VentureBeat

VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact.

Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:

  • up-to-date information on the subjects of interest to you
  • our newsletters
  • gated thought-leader content and discounted access to our prized events, such as Transform 2021: Learn More
  • networking features, and more

Become a member

 Source link

Back to top button
SoundCloud To Mp3