What is NOT AI?
by John Sumser, founder, principal analyst, and editor-in-chief of the HRExaminer
I’d bet you are buried in marketing claims and counter claims about AI in Recruiting and HR Technology. It is hard to sift through those claims to find the truth.
Artificial Intelligence (AI) is a precise technical term for a developing technology. These are the early days. The technology is in a primitive form. The long-term vision is to make conscious machines with human capabilities (or more).
We are not even vaguely close. The state-of-the-art tools can learn in a specialized way (see What is AI?). The tools can accept new information that adjusts the way they work. Using them involves sophisticated math to make a model of the world.
If you ask trained professionals about the many things that claim to have embedded AI, they shriek in horror. The closer you get to the technology, the clearer the difference becomes. Like all great engineers, AI technologists and data scientists demand precision in language and execution.
Artificial Intelligence is also a marketing term. Since no one regulates Marketing terms, they are rarely precise. They often grow, like a weed, into nearby areas. From a marketing view, AI is anything from super-complicated Data Science to scripted automation of existing processes.
Marketing language works to make you ask questions. It is imprecise by nature. Asking questions makes you more likely to engage with a product or service. Marketing uses words in ways that add to their ‘stickiness’.
There is a big ongoing argument about whether this or that product is really AI. Think of this as a debate between engineers and marketers. For the engineers, there is a true answer. In marketing, the standard is more like ‘generally true’ or ‘close enough.’
In other words, no one really wins those debates.
Marketers are not trying to fool you. Their job is to help make the product or service more attractive to the people who need it. Both marketing and sales people have a roomier definition of technical terms than tech professionals do.
Questions to ask before you buy AI
When the dust settles, it doesn’t matter whether something is or isn’t AI. What does matter is the value you get from a product or service. If a manual process can save or make you more money than an AI competitor, you should not buy the AI. If an AI-free offering gives you better results than its AI competitor, buy the thing without AI.
It is important to know what value to expect from a product or service. Understanding how it works and how to maintain it are questions that only matter if you plan to improve or fix it. If these are things the vendor does, why would you care?
With that in mind, consider asking your vendor the following things:
- Exactly what is the value my company gets from your product? How do I measure and verify that I am getting the value?
- How do I tell if it’s working?
- How do I tell if it’s broken?
- Are the results repeatable?
- Are there indicators that tell me when something changed?
- What are your plans for improving this product or service?
- Will you keep me in the loop when you change and improve the tool? How will I know?
- What training do you provide?
- Should I take the system’s suggestions and recommendations as an input to decision making or as the actual decision?
- How has your ethics board been involved in your design process?
That should start the conversation on a better footing than arguing about whether something is or isn’t AI. If you assume that most of what you hear about AI is marketing language, you will be better prepared for that conversation. The most important thing is not how it works but what value you get.