How AI and Machine Learning Get Smarter Every Day
Dataminr’s Alejandro Jaimes on the benefits of AI
Consider your search history and corresponding ads that pop up in your online profiles and Amazon cart. Or, how colors and lighting are automatically adjusted in photos to flatter skin and landscapes differently. Consider the convenience of taking a picture of your credit card to make an online purchase, rather than tediously typing in the numbers. This same technology customizes your online dating experience and social media profiles. Ideally, these machines have been taught how to make your life a little more convenient and a little more beautiful.
While the terms machine learning and artificial intelligence (AI) have become interchangeable, Alejandro Jaimes is careful to point out that they are technically different categories. “Artificial intelligence is a broad field that includes many sub-fields, one of which is machine learning,” says Jaimes, who is the senior vice president of AI and data science at Dataminr. “The ultimate goal of AI is to have computers behave like humans—to be intelligent as humans and indistinguishable from humans.”
While the utopian vision of AI is still far in the future, machine learning has been around for decades and is being utilized more each day. This kind of “intelligence” is dependent on data, which must be collected and labeled in an infrastructure that allows for machines to detect patterns. Machines build models from the patterns they detect, and later use those models to predict outcomes for the new data they come across.
Opportunities for machine learning seem limitless. While some industries (any that collect data and are not heavily regulated) are a natural fit for early adoption, Jaimes is confident that it can and will impact all fields. “AI can be used to optimize any task that’s repetitive,” he explains. “The more data, the more patterns, and the more AI can be applied to make better decisions and be more effective.”
Manufacturing, inventory management, and online advertising are easy examples of this, but machine learning can also be used in a small shop, for example, to determine prices. “Prices in a supermarket could be based on collected data,” Jaimes says. “Many Japanese grocery stores already use digital tags so they can change prices based on time of day or number of items purchased.”
On the spectrum of readiness for this technology, healthcare is one example of an industry that is at the beginning. Care providers already collect lots of patient data, but instead of using a machine to detect patterns and make a diagnosis, appointments are still scheduled (and paid) for highly common things like a cold or flu.
“Imagine if you could fill out an online form, use speech recognition to answer a few questions, and generate an automatic determination without a physical exam,” Jaimes says. “In healthcare, technology is often thought of as a cost instead of an investment, but a machine can learn much more and much faster than a human can in a lifetime, giving doctors more time to focus on difficult cases and on the human connection. It’s not about replacing doctors.”
And these advances are not far. The world is experiencing the biggest technology change that humanity has ever seen—change that will increase as hardware becomes cheaper and processing becomes faster. Self-driving cars, device interaction in the home, and voice recognition will become more commonplace as their technology improves. “New data, new sensors, and new opportunities to collect and apply data is happening rapidly right now,” Jaimes says. “Everything will be transformed because of it.”
These predictions may seem too fast or even unsettling to some, and Jaimes acknowledges that some concerns are valid and should be taken into consideration. “While it’s true that some repetitive jobs may be lost to AI, what we’ve seen so far is that if technology eliminates some jobs, it creates new ones,” he says. “We should focus on educating our work force to be adaptable for transition.” After all, machine learning is dependent on humans to properly label data.
Jaimes additionally encourages all consumers to be conscious of how data is collected and used. As AI becomes more integrated into daily life, it will have a bigger impact on human decisions, which is why awareness of algorithms is becoming more essential. “Knowing that an algorithm provides our top results in a search engine might inspire us to scroll past the first page,” says Jaimes, who also recommends popping any “bias bubbles” by avoiding filters in news feeds and social media. “Because what we’re fed by algorithms can be based on our own biases, it’s crucial to understand that not everything that algorithms show us is absolute truth.”
According to Jaimes, the most successful technology anywhere in the world considers human needs first. Cultural context, abilities, and Maslow’s hierarchy of needs, all need to be included in successful development of AI—and that applies to algorithms and data as well. “If we do that, I can imagine children in rural or dangerous areas, able to learn at home without fear of trekking to school,” Jaimes says. Similarly, if businesses are more efficient, they can have better prices, which translates to better access. Easier access to information is especially significant in developing countries. “The capability to improve lives at every level around the world is incredible.”
Boiled down to basics, the best AI will not focus on people as users who engage technology in a vacuum. Instead, it will depend on user-centered design that understands people as human beings. “A human-centered approach says, let’s understand the whole person, even beyond how they use the technology,” says Jaimes, who recognizes that what happens before and after using an app will influence a user experience.
If AI remains focused on the human experience, it certainly has the potential to make life a little more convenient, a little more beautiful, and make the whole world a better place.