Some people live and breathe AI, while others can’t differentiate it from a comment by the Fonz… “Aaayy!” Artificial Intelligence has been around for a long time, at least since the birth of Eliza at MIT in the 1960’s. Market trends reported by the organizers of AI World indicate that AI is rapidly growing into a multi-billion dollar market, and this year’s AI World showed a plethora of companies talking about computer vision, machine learning and everything in between.
So, what is AI and will it take over the world, turning humans into slaves and toys to the robots? If science fiction were true, we might see things like this, but that’s not the case in the real world. Let’s explore a bit of what vendors had to say at the Expo.
The Expo Hall was more intimate than at most conferences. It was in an area of smaller, interconnecting rooms. As AI technology offerings advance, I’m sure the vendors will move into a much larger hall. Still, indicative that AI is coming of age is the presence of larger companies like Dell and Deloitte who will help to spread AI into the mainstream.
As someone who supports startups and connects them to potential partnering opportunities, I was most interested in those companies that had components that could enhance the value of a startup’s current offering or the companies that could provide a look into the future for vision planning. For this reason, I spent less time with companies targeting large enterprises.
As a note, AI is being applied to virtually every industry and market. Check out some of the research and analysis done by companies like CB Insights. Here’s their top 100 AI companies for 2018 (I did not see any of them at AI World).
Before we get into the demystification panel, let’s explore some of the technology found in the Expo area. I organized the concepts into some general categories, which may help with understanding.
Rapid Pattern Matching
Photo by Adrian Rosebrock
A popular use case for AI is to speed up pattern matching exponentially to learn about what something “normal” looks like. Two companies at AI World doing something similar are Alegion from Austin, TX and CrowdFlower from San Francisco. These companies have a platform where pattern identification tasks are outsourced over the Internet to people around the world who want to do the jobs for very little money. Then, the pattern identification is sent back to the AI engine to learn about what a normal version of the object looks like. For example, the workers see photos of car parts, draw lines around their boundaries and identify whether there are any defects. When the AI engine processes the data, it learns how to identify on its own whether a part is good or bad, whether it should be kept for production or sent back for repairs or scrap.
The interesting thing about this solution is that humans are helping machines learn about something which will automate away the human job in the future. Since there are so many applications for this today, it’s not an imminent concern; however, it is something to ponder. Both pitches were so similar, I could have closed my eyes and used my own pattern matching to determine they were the same company. They also have an added issue which is staying two steps ahead of their big competitor, Mechanical Turk, which is owned by Amazon. So far, they say they have much better solutions, but that can change quickly.
Improving Human Performance
Some companies are targeting ways to use AI to improve how people work. This generally means automating repetitive up-front tasks or making it easier to interact with complex systems.
For example, eXalt solutions from Boston will spend 6 weeks with a company implementing rules to provide customers with the initial sales support they needs, so a human sales rep is not needed. They showed me how a brief Q & A interaction with the AI bot would generate the full specs and pricing for an Oracle computer system. Once the customers know what they want, they contact the human sales rep to finalize questions and seal the deal. This is a great way to improve productivity, but there is an upfront cost and it may not work as well with all types of pre-sales interactions.
Interactions from Franklin, MA helps customer service call centers save money and improve customer service by using interactive voice response and AI to respond to many first-level customer questions that no longer need human assistance.
Startup product intrprtr from Sunnyvale, CA shortens the length of sales calls by making it easier for sales reps to get information out of Salesforce to answer questions and move a sale forward. In other words, when software like Salesforce evolves into a complex beast that takes too much time to learn to use effectively, AI can come to the rescue to make interaction easier – a simple UX for the human with detailed Salesforce rules built into the bot. Read about chatbots by founder Frazin Shahidi.
Better Pattern Matching and Recognition
Digging deeper than the first two categories, there are companies developing solutions that match and/or recognize patterns using more of a fuzzy logic. They look at many more data points and layers and try to identify matches of objects that are not exact matches. These engines require building in something that resembles intuition. They come closer to behaving like people do, in that they can make interpretations. This type of AI takes a lot of iteration within the learning process.
One company I spoke with is a job placement startup called JobRobin from NYC. They try to create a fuller profile of a worker in order to make a better match to a job. Their questioning process includes softer skills questions and AI to interpret personality traits based on work data, skills data and the data from the softer skill questions. Their goal is to find better matches for jobs, even when the work experience may not be an exact match. They are currently focused only on data professionals. I’m intrigued by what JobRobin is working on, since it fits into a Startup Weekend project I co-founded on how to find jobs for out-of-work blue collar workers in middle America. We designed a system called “ReSkillMe” that has some similarities to what JobRobin is working on. All the best to them in this endeavor!
Enhancing Education
I had a brief conversation with people from the University of New Hampshire, where they are leading higher ed in online learning and competency-based learning. Our futuristic thoughts were about whether AI could enhance online learning, which is often not as engaging as in-person classes. The possibility of creating a TeacherBot is very interesting, but alas, I did not find anyone working on this at AI World.
Learning About Humans
Parna under the spell of the Affectiva AI software
Humans are very complex and out complexity puts AI in its place as automation technology and not humanized technology. Because we are illogical, we are difficult to understand, and our actions can be difficult to interpret. Using AI to better understand human behavior has got to be much harder than just recognizing patterns, even fuzzy ones. At AI World, I did not find anyone working on human emotional learning, but there is a company working on recognizing human emotions through facial expressions and body language. The company is Affectiva out of Waltham, MA. Using computer vision, the software looks at multiple data points on a person’s face and then interprets emotions. They’ve been at this for a while, so they have learned a lot about how to interpret this data. A demonstration of Parna’s expressions during the panel discussion was pretty cool. A smiley face or other emoji showed up next to her face as she changed her expressions.
Now that we have wandered through the Expo area and touched on some of the ways AI is being used today, should we be worried about what all this Artificial Intelligence will do to us? Will the robots win? What do you think?
Read the next post to hear what a panel of experts thinks.
Now that you may have a little better insight into AI, explore more and see how you can apply AI or other new technology ideas to projects you and your colleagues are working on. I’d love to hear your stories of discovery, experimentation, success and failure. Get in touch at ebraun@ssinnovation.com or on twitter.