(This article first appeared in Forbes.)
In a crowded, rapidly developing field like artificial intelligence, decision makers in enterprises can struggle when deciding where to invest their company’s resources. Almost every emerging company providing data science services will tout their AI expertise, but which vendors actually have the proficiency required to deliver valuable change and which are only claiming they have AI capabilities to capitalize on the buzzword?
The answer does not lie in the pedigree of funding or press surrounding a company. While these are important, both are all too easy to come by in the current environment. To determine if a company is truly a worthwhile partner in artificial intelligence (AI), here are five questions to ask.
1. Why Did You Get Started In AI?
Following the framework of Simon Sinek, start with the why of the company. This response should indicate if a company has deep-seated motivation to contribute to the field or if AI is an afterthought tacked on to existing products. These days, newer companies that have built a business around what can be done with a product are at an advantage over older companies that will have to make radical adjustments to their strategy to incorporate technology like AI.
For example, Tesla considered the self-driving feature a fundamental part of the driving experience, a forward-thinking mindset that is reflected in the construction of its cars and puts the company ahead of those struggling to add the feature now.
Why a company got started in AI should almost certainly be part of its message, but it should also be evident in the time and resources the company dedicates to research (which, given the rapid progress of the field, is a requirement to deliver efficient solutions).
2. How Much Do You Know About AI?
Attracting the scarce qualified talent in the field of artificial intelligence is extremely difficult. Element AI has estimated there are fewer than 10,000 people in the world who are qualified to tackle serious AI research. The majority of these experts either stay in academia or head to giants like Google or Amazon, so there are a precious but motivated few working at AI companies. Businesses that are able to attract this talent are thus doubly appealing.
Ask what the company looks for when hiring data scientists, and watch for indicators like the number of patents and the number of employees with qualified advanced degrees to determine the quality of the team. Most importantly, come prepared with questions (even something as simple as “Can you explain how you use deep learning?”) to zero in on technical knowledge. Understand the depth and the breadth of the company’s capabilities beyond commonly known approaches.
3. How Many Deployments Do You Have?
Go beyond hype surrounding a company: What has it done? Does the AI vendor have the stability to work on long projects with enterprise clients? Many fledgling AI companies complete proof-of-concepts but never move beyond that stage or come up with a product strategy or services strategy.
Use cases should show clear, practical business value, and the company should be able to provide referenceable accounts. Scaling AI and ML solutions is not trivial, so challenge vendors to share use cases or plans to scale their service.
4. What Is Your Delivery Model For Projects?
Preparing a company for AI technology is a difficult task, and actually implementing the solutions is even more challenging. Thus, the relationship with a vendor should not be simply transactional but a consultative one. It’s also worth considering what kind of company is needed for your use case. A service company can help you solve a problem once, whereas a product company will give you the ability to solve multiple problems.
Another indicator of a mature delivery model is a strong partner ecosystem to deliver an end-to-end solution. It means the company has attracted the attention and trust of other companies in the space and ensures the vendor can focus on its own
5. Do You Have Domain Expertise In Our Field?
One of the advantages of AI is that it can deliver a different way of looking at a problem. A client of ours likes to recount the story of a nerdy data scientist who had never seen an oil pump before explaining to a room full of good ol’ boys from Texas the new data-driven model for when a pump was about to break. Turns out, her method worked better than their reliance on their eyes and ears.
However, having subject matter experts on hand who understand the application of data science is critical to project success. The vendor should be able to bridge the gap between data models and implementation. With domain experts facilitating a collaborative process, this ensures positive change management and a higher rate of adoption and success for the project.
Though artificial intelligence is at the top of the hype cycle, executives have pressure to show return on an investment in the new technologies. Choosing a strong partner will ensure a project demonstrates why there is so much buzz surrounding artificial intelligence.