The Power of Vertical AI in a Monolithic AI World
During South by Southwest, our CEO Amir Husain spoke on a panel to ponder “The Power of Vertical AI in a Monolithic AI World.” Joined by John Price of Vast.com, Khari Johnson of VentureBeat, and Mary Ann Azevedo of CrunchBase News, the group covered the practical uses of AI, its risks, and the future of the field.
Since not everyone could make the trip to Austin for the festival (and see Amir representing his alma mater in a burnt orange tie), here are 10 interesting takeaways from the panel.
1.) “If you’re at war in the South China sea, you can’t send that query to the cloud and wait for a response.”
Husain brought up this point while discussing the advantages of narrow AI (i.e. AI with a specific purpose and meaningful economic outcome rather than generally intelligent AI). He argued that all the investment going into AI today will be paid back “manyfold” when AI can diagnose cancer better than a human, when an autonomous car can control a car better than a human, when a weapons system can operate better than a human.
2.) “It’s a very steep curve, and if you decide to get on it, you have to stay on it and keep running.”
Price brought up that the field of AI is young and constantly changing, emphasizing that rather than focusing on what humans are capable of and how AI can help, he has to look ahead and think about how the next version of a software may put the current product out of business.
3.) “There’s very interesting AI out there that nobody is using because it requires a good, no, a great product person.”
Though the technology is fascinating, user interface and design still play a big role, points out Price. People have to be able to make sense of the data being crawled, and interactions with the complex system should be understandable.
4.) “It’s not the case that those domain experts are required to prescribe a solution.”
Husain described how SparkCognition’s automated model building (AMB) technology enables work in many different verticals with only a light touch of expertise. While experts are still required to listen to the client, understand the problem, and act as “translators,” AMB produces data-driven solutions quickly without the need for background knowledge.
5.) “When data isn’t just what movies I like, but multi-billion dollar test equipment, and handing that data off to someone has connotations.”
Having worked with industrial companies, Husain understands that businesses can be protective of their data. Thus, the ability to make decisions at the edge is key (not to mention, companies need to think about cybersecurity).
6.) “Do you want to pull data from your database and see a fancy chart that says this might be an anomaly? Come on, that’s so 1979.”
Husain acknowledged most AI capabilities are currently based on the technology’s ability to perceive, which he thinks is only scratching the surface of what it can do. He is interested in AI that goes from observation to decision to simulation to action, and how to reason and enable action that’s explainable.
7.) “We’re doing ourselves a disservice by pretending that AI is not going to change the entire landscape of human value.”
Price told the story of meeting with a domain expert who had six rules for solving problems that he had created a career out of, and he “wasn’t excited about a little punk getting these rules and putting them into software.” It’s possible to automate extraordinarily complex tasks, and the fears of job loss are prevalent.
8.) “We can’t just build this tech and not revamp our communities.”
Price and Husain agree that the implementation of AI requires thoughtful discourse on change in aspects of society like politics and education (regarding the latter, Price quipped, “One thing’s for sure: [children] won’t have to learn penmanship. They certainly won’t have to memorize anything.”)
9.) What are good areas to go into with AI?
A final question from the audience elicited interesting responses from the panelists:
“Don’t do anything that requires you to sell to a bank. You can’t live long enough to get them to change anything.” – John Price
“Go find a problem that really bugs you and try to solve it. Because chances are, it bugs a lot of other people too.” – Amir Husain