AI vs. Machine Learning: The Costs of Innovation
Artificial intelligence (AI) has been making serious technical progress over the last several years, but not in the political sense. Tech giants like Microsoft and Google, and online retailers like Amazon, have found new ways to accelerate their products using AI-driven algorithms. AI isn’t exactly the correct term, however — at least not in the sense that general consumers know it, in regard to machines like HAL from 2001: A Space Odyssey or Skynet in the Terminator movies. There’s a difference between AI and Machine Learning, but it’s such a new concept for the zeitgeist, the two are easily confused.
AI is a broad term encompassing technology that employs advanced computer intelligence, but it’s Machine Learning (ML) that really gives computers that human-like intellect seen in science fiction. The difference between AI and MLmay be subtle but it’s still substantial enough to merit its own classification in the technology field. Whereas AI uses preprogrammed, rule-based information to perform specific actions like IBM Deep Blue defeating chess grandmaster Garry Kasparov in 1996, ML can, instead, use large amounts of data to teach itself how to do a task. Another example, Google’s DeepMindlearned how to play the Japanese game Go against itself and then went on to defeat South Korean Go player Lee Sedol in 2016.
It’s not so surprising that many rational people are daunted or even haunted by the very concepts behind AI, because as humans we’re often rattled by things we don’t yet understand. What we have to accept is that our notion of privacy will change over time, as will our relationship to technology. Worst case scenarios aside, nothing has been lost. Our relationships to these technologies is ever-evolving, net neutrality or otherwise. The world continues to spin, time continues to pass.
We still have the power to decide how we relate to the internet and the silicon creatures we connect to it. We can be good people who get online or bad people who get online. The online component is not where morality comes into play, it starts from within. There is no reason to fear AI more or less than anything else (or ourselves). The people alive today do not own the monopoly on the fear or sacrifice that comes with unleashing AI to the world. We’re also not unique. Early adopters bear the brunt of inconvenience. People in Victorian times depended on candle lighting and little else about their lives compares to ours now. The Victorians were once afraid, too. They were afraid of many things we are not, like getting help recovering from routine illnesses, giving birth without complication or becoming an outcast by high society for having differing opinions or lifestyles. Victorians were also distressed about installing gas lighting in their homes and rightfully so. Unfortunately, along with progress came great human decrement; fires, explosions and all sorts of breathing issues.
“Gas must have provided a quite stunning improvement to people’s ability to read, write or sew in the evenings with minimal effort. It nevertheless had many drawbacks. There were frequent explosions, and it replaced the oxygen in the air with black and noxious deposits. The aspidistra, a hugely popular plant, became so because it survived well in oxygen-starved conditions. Victorian ladies frequently fainted partly because of tight-lacing, but also because of a lack of oxygen in their gas-lit drawing rooms.” -Lucy Worsely
There’s always going to be a personal tariff paid for the advancement of technology. For 20 years I’ve carried a cell phone that the state of California recently saw fit to warn us about the dangers of. I have arthritis in my right hand from too much texting (OK, tweeting) and my eyes have likely suffered from all the screen time, but this is part and parcel of what comes with the betterment for homo sapiens.
Now is the time we as a society get to decide how we want to allow AI to enter our lives. We can decide upfront that AI capable drones will never have authority to kill humans or that it’s illegal for AI robots to raise your children or stalk you or any of the freaky things we’re going to see.
What might creep you out is that ML allows online retail and social media sites to learn about their users on a deeper level. Every tweet, Facebook like, and Amazon wish-list item is filed away on servers that have algorithms sifting through the data to create an online profile for each user, which then allows companies to push specific ads or relevant digital content like Youtube videos. Yelp uses ML to sort massive quantities of user-uploaded pictures into easily consumable categories for users. Pinterest’s ML algorithm learns a user’s preferences and search patterns in order to suggest other images and content that the user would like. Amazon also uses ML to drive sales by suggesting items similar to what a customer has in their search history, previous purchase record, or in their shopping cart.
Another super popular application for ML has been Chatbots, which uses Natural Language Processing (NLP) to learn how to interact and create human-like conversations. Staples teamed up with IBM Watson to transform their Easy Button into an ordering system. With IBM’s Watson Conversation system, NLP, and ML, the Easy System is able to interpret a customer’s request and provide the correct response based on the information available. To an extent, Google’s “RankBrain”, a part of its search engine algorithm, also operates in the same capacity. Google’s search engine takes in a user’s search parameters and sorts through the Internet to return results that match their inquiry. Users demand relevance so search must always find it. The only way that relevance can be achieved dynamically and randomly on-demand is with AI and there’s a fair bit of creep walking down that path, but all we have to do is open our eyes.
Now that we understand how AI and ML can help us, let’s discuss why we feel malaise about AI. Because an unknown entity that makes us afraid is often just a foil for a dangerous ‘other.’ It turns AI into a fear of ourselves, because we are uncertain of where we’ll lead it. We are not condemned to dystopian futures of robo-overlords unless we don’t agree to ethical standards and everyone goes all weird and rogue. Quite a bit of data science is still manual and involves many humans, which means we’re not exactly at scale. The autonomy of machines is greatly oversold in movies and TV. Plus I have to ask overall; why would machines or robots want to hurt us?
We are never going to be slaves to our devices, it’s quite the opposite. AI is not a deity, it’s a constructed foil. We’re afraid of losing power in terms of sentient beings, but I really don’t see that as an issue:
-We’re not as advanced as we think we are. The practice of data science is still largely human and therefore subject to its’ bias.
-Evil intent comes from evil humans, not technology. We should be afraid of the humans behind the scenes, if we must be scared, not the technology itself.
-Fear holds us back. I will not be held back.
Look at the potential benefits, there are too many to count. Brains and faces can be replaced, hearts replaced. Yes we’re stumbling poorly between the light and dark of morality on a public stage and parsing the wash of human emotion over the internet. That’s OK. We’re also seeing a generation who could easily have ten more years of their parents being around because they have tools to get them healthier; step trackers, heart monitors, nutritional support, etc. Diabetics won’t have to remember to prick themselves, their eyes will tell their phone to tell their bracelet to medicate them, etc.
Everyday, ML bridges the gap between science fiction and reality. Its continued evolution promises to bring exciting changes to both online and real-world interaction. It’s only a matter of time before systems like Siri and Google are able to hold a full conversation with users that is nearly indistinguishable from human communication.