How to Spot an Artificial Intelligence/Machine Learning Business Opportunity
You’ve likely heard key business phrases like artificial intelligence, machine learning, deep learning, computer vision, neural networks, adversarial neural networks, and the like. These phrases are all the rage right now, and for good reason since these technologies are revolutionizing business and industry. So what if you want to jump on the bandwagon and introduce this new tech to your field? You’ve always had an entrepreneurial streak in you, and now you’re ready to act on the right opportunity, but how do you know what the “right opportunity” is? Here are some things to consider as you mull over your opportunities.
Identify needs in an area you know well—or your cofounder knows well
First, you’ll want some domain knowledge in one field or another to spot a good application for this tech. The more you know about an industry or area, the more you’ll be able to identify what in that sector could benefit. It’ll be more difficult to break into an area where you know very little. That’s not to say it’s impossible to move into a space you’re not familiar with, it’s just more difficult.
The more you know about an industry or area, the more you’ll be able to identify what in that sector could benefit. It’ll be more difficult to break into an area where you know very little.
Here’s an example of someone who developed AI/ML tech in her industry by leveraging her domain knowledge. Meet Alexa Anthony of Magic AI. She competed in horse-riding competitions, and during that time, she experienced a tragic loss. Her horse contracted a common horse ailment called colic, a disease that quickly claims its victims. A horse’s life can be saved if someone identifies the symptoms early enough, but that window is short. Unfortunately, Alexa didn’t find out her horse was sick until it was too late. After experiencing this deep loss, she spotted an opportunity to use computer vision to help other horse owners avoid the same fate.
Since she already knew about AI and ML tech options, she saw an opportunity to design a machine-learning application that could identify classic colic symptoms in horses before it was too late. The system, StableGuard, watches horse behavior in the stable and identifies and pings a horse’s owner if the system identifies unusual horse behavior. Without leaving their houses or dinner appointments, owners can access the horse cam on their phones and see the flagged behaviors themselves. This gives them the power to call a horse doctor and get help right away if they agree that the behavior is strange and needs to be assessed by a professional.
Now could an outsider have identified this issue? Possibly, but Alexa had the large advantage of knowing this was a common problem and of personally experiencing the pain of losing a horse, so she was convinced the tech was valuable, and she was passionate about the cause. So if you have passion about fixing a problem and domain knowledge about the sector, consider this a green flag.
Identify the size of the problem
While we’re on the farm, here’s another example that illustrate a slightly different point. Matthew Rooda grew up on a pig farm and eventually managed pig-farm operations. He knew the industry well and had experienced firsthand how frustrating it was to lose piglets—many piglet deaths happen when their mothers roll on top of them, accidentally crushing them to death. He knew if he was experiencing this at his farm, so were lots of other pig farmers. Motivated by this, he started a company called SwineTech to address the problem. He build a belt that buzzed the mother (kind of like a dog collar, but less intense of a shock) whenever the equipment identified a distressed squeal from a piglet near the mother. The buzz from the belt would prompt the mother to move and often save the life of a piglet or two.
The reason why Matthew’s company could be successful, however, wasn’t just that this was a problem and that he knew about it. It’s that it was a large enough problem. There were huge potential profit margins for an innovative solution. Each piglet is worth about $45, and every year, the industry loses about five percent of piglets in this way, so saving some portion of five percent of an industry’s potential profits offers large profit margins.
Know the strengths of AI/ML
While you need to have domain knowledge in an industry, you also need to know what artificial intelligence and machine learning can do well. Ask yourself the following questions to evaluate its potential in your field:
Does your industry or could your industry use photographs to identify something, quantify something, or create something? If it does or can, there may be an opportunity.
There are lots of examples where innovators use AI/ML to identify faces, objects, animals, words, shapes, patterns, and colors in pictures. The medical field uses AI and photos to identify cancers and discover new brain regions; the conservation industry uses the tech to identify tree density by counting number of treetops in photos and even identify animals caught on camera-trap photos in the wild; policing entities are using it to identify faces of suspects in a crowd, flag child-abuse images, and map individual faces over facial heatmap data.
Other applications include using AI to almost imperceptibly change photos so facial recognition can’t identify the faces in them, to create faces of people who don’t exist, to fill in pictures that have gaps and holes in them, and to form 3D objects from 2D ones. Here’s a more complete list of what you can do with photos and AI/ML:
- Picture recognition
- Text, object, facial, and pattern recognition
- Picture generation
- Photo creation from text or from other photos
- 3D image creation from 2D images
- Picture adjustment/enhancement
- Image stitching
- Image extraction
- Photo imitation (make any photo look like Van Gogh painted it)
- Hole filling
- Facial recognition blocking (changes image to block facial recognition)
Very similar to the photo category is video. Does your industry use or could it use video to meet its aims? If it does or can, there may be a business opportunity here.
The entertainment industry can manipulate facial expressions on a character, change the face of one person to another in a video, and change one aspect of a character’s face, like deleting a mustache; the surveillance industry conducts searches for things in videos—like a Google search would for images, potentially identify specific behaviors in video, like violence, and potentially identify someone by the way they walk; the industrial sector can identify problems with hard-to-access assets by using AI with drone video; the conservation industry can count how many fish climb up a fish ladder while simultaneously identifying fish species, scan and sort recyclable trash using computer vision, and identify plastic trash on beaches. The list of AI applications could go on and on and on. Here’s a more complete list of what you can do with AI/ML and videos:
- Video object detection
- Identification of behaviors, movements, or objects across many videos
- Facial recognition
- Video manipulation
- Facial control of character in a video
- Face swapping
- Video generation
- Video creation from other videos
Does your industry use or could it use audio to meet its aims? If it does or can, there may be a business opportunity here.
The agriculture industry uses chicken chatter to identify when chickens are too hot, otherwise distressed, or sick; the maternity industry attempts to identify what your baby is expressing through its cries; the conservation industry identifies when chainsaws and logging trucks are illegally deforesting a protected area.
Other applications include telling Siri to dictate a text, taking a one-minute snippet of someone’s voice and recreating their “digital” voice, and commanding Alexa to do something. Here’s a more comprehensive list of applications using sounds:
- Sound recognition/sound isolation/speech to text
- Advanced language translation
- Voice isolation in a crowd
- Sound classification and identification
- Sound generation
- Non-language communication sounds
- Voice or sound imitation
- Music creation
- Natural speech generation
Undoubtedly your industry uses text in one way or another. Is there a way you could capitalize on this?
The law industry uses AI/ML to help create and analyze contracts; in language translation, Google uses it to help with language translation accuracy; the news industry uses it to write news articles; and the graphic design industry can use it to design pictures from captions. Here’s a more comprehensive list of what it can do in this realm:
- Text recognition
- Handwriting translation (from handwriting to computer font)
- Natural Language Understanding
- Document classification
- Topic modeling
- Picture creation from captions
- Sentiment analysis
- Language translation
- Natural Language Generation (still not very sophisticated)
- Article creation
Recommendation systems, targeting, and personalization
Most industries can benefit from advanced targeting and recommendation systems. Is there an area in your industry that could benefit from implementing something like this?
In the entertainment industry, these systems recommend movies, music, and the like, think Netflix, Spotify, and others; in the fashion industry, they can recommend clothing for an occasion (wedding), stage of life (maternity), or holiday (Halloween); in the job industry, they can better find candidates for a job; in the retail industry, they can offer product recommendations (think Amazon and Target) related to your purchase; in the ad industry, they can more precisely target audience identification for ad placement.
Mixing up different types of AI/ML
Many business applications of AI/ML splice together multiple different types of the tech to form a product. Take for instance applying soundscape to pictures. Step one is identifying objects or elements of the video feed, and step two might be a recommendation system suggesting a song for videos that have certain objects or elements in them. Could your industry benefit from a tech combination?
Other use cases
There are several other areas where businesses can use AI/ML to create valuable products, like wireless sensors that send data through the Internet of Things (IoT) to be analyzed by AI/ML for key insights. One example of this is sensors used with wind turbines to improve efficiency. AI/ML and IoT opens up a huge realm of possibilities.
Another unique area is using this tech in anomaly detection, where AI/ML combs through massive amounts of data looking for unusual activity, like fraud. This technology is able to identify unusual behavior, flagging it for an authority to review.
Identifying an opportunity is just the first of many steps toward developing a business. There are several steps beyond this that are necessary in order to develop a good business, like researching if someone else has already done what you’re thinking of doing and identifying if people will actually pay for your product. But this is a good start.
Hopefully the use cases and outlines in this article spur your creative juices. There are undoubtedly a lot of business possibilities in your industry. If you want to peruse even more applications of artificial intelligence and machine learning, we are cataloging unique AI applications in an ever-growing database that you can browse to get an even better idea of the possibilities. Also, get the latest information about cutting-edge applications of this tech by signing up for our biweekly newsletter, which reviews the top new applications every two weeks. To hear from some AI/ML creators, listen to our podcast, where we keep our audience up to date on the AI/ML applications and developments.
And if you’re really ready to jump in and create a product but are looking for some help, give us a call, 801-369-1702, or email us, [email protected]