Workiva: Balancing the Demands of GRC and ESG
What should enterprises consider when choosing an AI vendor?
According to IDC, worldwide spending on artificial intelligence (AI) could reach up to $58 billion by 2021. Although many business leaders are striving to invest in AI, choosing an AI vendor can be challenging. In order to address this, software company DataRobot has produced an insightful whitepaper with five key things to consider when choosing an AI vendor. The report also details the top challenges businesses face when evaluating vendors.
Choosing an AI vendor is complex
While many enterprises are seeking to invest in AI, selecting a vendor can be incredibly challenging. Indeed, the report states that "hype is inherent in new technologies, and AI is no exception." In fact, many of the vendors that tout AI do not actually deliver the technology. For example, many database vendors claim to offer AI but "since you need data for AI, they say, a data platform is an AI platform." Moreover, some legacy software vendors add AI functions to an ageing code base and insist that it is in fact AI. Nevertheless, it is evident that "legacy software cannot support AI at enterprise scale." In addition to this, it is common to rebrand software as "AI" without changing the product. Finally, the report insists that some vendors claim to offer "AI in a bottle" but in reality they "employ data scientists behind the scenes who do the work."
What to look for when choosing an AI vendor
As the aforementioned challenges demonstrates, transforming a business with AI is not easy. Nevertheless, DataRobot outlines five key things to consider in order to alleviate some of the issues associated with choosing an AI vendor.
A pragmatic approach
First of all, it is integral to looking for vendors who strike the right balance between theory and practical experience. While many AI theories "look great on paper," they do not always work in practice.
DataRobot also recommends seeking a prospective vendor in Crunchbase, a platform for finding business information about private and public companies. Moreover, report proposes to take "extra care when evaluating companies that are less than three years old" as it takes that long for a startup to take a product to market, acquire customers, and become organised for growth.
Modern software engineering
Next, it is crucial to look for an AI platform that is scalable, elastic, and has the ability to support the needs of large enterprises. Look for AI vendors that build on a foundation of open source software, develop an AI solution on commodity hardware, and choose a platform that is simple to deploy and manage.
Support for a broad range of users
"It’s not enough to have many different client applications," according to the report. Enterprises must therefore seek a vendor that offers open APIs and flexible business partnerships.
Support for the AI lifecycle
According to Gartner, it takes an average of three months to deploy a model. Defining a process for AI "from data through deployment is the key to reducing cycle time," so it is crucial to choose an AI vendor that understands and supports the complete lifecycle.
Check out our latest podcast, in which technology influencer Sarah-Jayne Gratton discusses removing the confusion around AI