In this special guest feature, Rick Nucci, CEO of Guru, provides a perspective targeted at people broadly in tech who know they need to start thinking about AI, given it’s invariably affecting their industry, but don’t know where to start. He provides expert advice on how they should go about wading through the hype and choosing an AI vendor. Rick has more than twenty years of experience in creating category-leading software solutions and companies. Prior to Guru, Rick was the founder and chief technology officer of Boomi, which defined and led a new segment as the first ever cloud integration platform-as-a-service. Boomi was acquired by Dell in 2010, where Rick went on to run the Boomi business for Dell as its general manager, helping grow the organization into the industry leader it is today. He holds a Bachelor of Science in Logistics, Materials, and Supply Chain Management from Penn State University.
Artificial Intelligence is a fascinating technology that has had more ups and downs than anything else in the field of computer science. So much so that we have seasons now for AI! When things are going well, we are in an AI Spring, and when things are not going so well, AI Winter. We are currently in an AI Spring, mostly due to cloud-based architectures that allow for unparalleled computing power and data storage — two things AI requires to be successful.
With this “Spring” has come a huge hype cycle, with every vendor touting some form of AI as a differentiator. None of us want to miss out on the enormous potential, but we also need to make sure we’re taking very calculated risks so we don’t end up with snake oil or worse.
How can we find balance somewhere among the disillusionment and value-add? Who can we trust? Asking the right questions is a great first place to start.
What to do in this world of hype
There are a few key asks that I’ve found are universally useful for evaluating legitimacy in the AI industry. These are built to help you pick solutions that will have a positive impact on your organization, while pretty quickly weeding out the chaff.
- What metrics should we expect your solution to improve? We are (at least) decades away from the idea of Artificial General Intelligence, something that approaches a true multi discipline, self learning machine. Successful AI today is very focused on specific problems, so beware the jack of all trades here, they master nothing. Cut to the heart of the problem by first asking vendors what metrics you can expect their solution to improve. This results-oriented approach serves a dual purpose of zeroing in on where a vendor can (or can’t) actually help you, and allowing you to quickly separate out the “jacks” from the ones who will do the thing you actually need them to do.
- How does it work? This sounds painfully obvious, and I guarantee it’s already a question you were planning to ask. However, it’s the answer that is most important to focus on here. Many AI solutions like to lead with their algorithms, describing which techniques they use, why they are the best ever, and why it’s the only way to solve the problem. This would be like if you were shopping for a car, and the first thing the sales person did was show you the schematic of the car and how the engine, transmission, and hydraulic systems work. It’s not about the algorithm! You should expect to hear a very simple and clear answer to this question, with a strong focus on the end user’s experience with the solution. It may be very complex “under the hood”, but your project will not be successful if it is creating additive or complicated work for your team in order to get value from it.
- How does it learn? Much like the last enterprise software movement of cloud computing, trust and transparency are critical. In the world of AI, the technology “learns” by absorbing large amounts of training data in order for it to be successful. But what exact data does the AI provider need? What is being accessed and stored? How long is it stored? Asking questions about how it learns will uncover these critical answers. If vendors can’t be crystal clear and direct about exactly what data they gather and why, then run.
Once you’ve established that the solution (1) will get you results, (2) works well, and (3) takes data incredibly seriously, it’s time for my favorite question (and one that will probably bring sighs of relief to workers everywhere): How will this solution make my team better at their jobs?
There is a huge focus on the automation aspect of AI, the idea that because of AI, jobs will be eliminated and humans will be replaced. Someday, this could happen, and it’s great that attention is focused on this and the potential impacts on society. But we are far away from this reality. Real, useful AI solutions are complementary and make teams better at their jobs, they don’t threaten to replace them. Vendors worth your time are the ones that will show you how their solution will boost your team, not eliminate the need for them.
The example I always like to use is customer support. This is an industry already adopting bots and AI-driven chat functionalities that exist to automate the interaction between your customer and support agent. But machines are far from understanding empathy and any customer support team will tell you it’s all about the human touch. If a customer is frustrated, contacts support, and is greeted with a robot, they quickly go from frustrated to furious. Let’s remember that as we think about where technologies like AI are actually best poised to help.
AI will the drive the next era of software, and be even more transformational than the move to cloud computing before it. While the hype is deafening, there are real gains to be made today as long we approach AI projects with our eyes wide open.
Sign up for the free insideBIGDATA newsletter.
Leave a Comment