There's a ton of hype out there about artificial intelligence.
It's easy to start thinking AI can do everything under the sun.
But, while AI is certainly impressive, it's not all magic.
It has limitations. And it's crucial that you know what the technology is—and isn't—capable of before you start researching solutions.
That way, you can honestly assess how AI tools fit into your current tech stack.
So we asked six marketing AI experts to go beyond the hype and share their perspectives on what you need to know about AI's real-world capabilities.
People think either apocalyptically (“The machines are coming!”) or quixotically (“Just imagine if AI could do this!”), but none of those perceptions solve actual business challenges.
Therefore, the lack of understanding of AI is the main limitation. For example: “AI” itself doesn’t really mean anything, but it’s a catchy buzzword. It’s a catch-all phrase for numerous application areas such as NLG, NLP, machine vision, machine learning, and so on.
- Parry Malm, CEO of Phrasee
Most use cases are generalized, such as “if you like apples, maybe you’ll also like pears” or “here are the possible ways a person might say ‘apple.’” With recent breakthroughs in computing capacity, deep learning has made it feasible to provide the system with all the possible inputs to questions like these and build a model that’s highly effective. However, even in “best case” systems like these, careful problem formulation is key and still requires significant resources to deploy.
Problems that do not lend themselves to a generalization, do not have a clear set of “known good criteria” to provide training inputs, or are simply sufficiently complex are still very expensive and difficult, if not impossible, to get good results from.
- Erik LaBlanca, CTO of Seventh Sense
These techniques are wonderful for many problems, including collecting insights from the market. But they can't solve everything: brands often demand more control over their content, and want to guide the conversation with the market in a certain direction. They also want to not just follow what's happened before, but also be different and get attention by standing out. Big data applications are good at following the crowd, not so good at striking a new path.
- Andrew Bredenkamp, CEO of Acrolinx
The process of collecting samples and classifying them is labor intensive and can be error prone.
- John Osborne, CTO of Crayon
Artificial narrow intelligence such as ours is excellent at pattern recognition and provides excellent intel. But to get great results, there's always a trained human in the machine.
- Aki Balogh, CEO of MarketMuse
<We believe the best applications of artificial intelligence today are those that tackle the challenges introduced by big data (across industries and applications) and make sense of it all, as well as those that automate recurring, manual tasks—but on a scale that can’t be achieved by humans.
But humans are unique in their ability to feel in a very complex way and translate those feelings into emotional connections.
Limitations in artificial intelligence will also stem from the degree of precision with which technologists are able to replicate human “intelligence” and decision-making.
- Or Shani, CEO of Albert