"Artificial intelligence" still has a sci-fi ring to it, but it's been on humanity’s agenda for six decades. That's how long it's taken for A.I. to enter our everyday lives in meaningful ways — and now it's here to stay.
In 1956, a group of researchers led by John McCarthy and Marvin Minsky created an "intelligent" machine that could solve word problems and speak English. Since then, A.I. research has seen many failures and success, but we are starting to see its potential more clearly in the form of chatbots. They have managed to infiltrate every sphere of life, including dating, politics, learning languages, and healthcare. They might even herald the second internet revolution and, eventually, the era of singularity.
Like people, A.I. chatbots have different personalities
Delving into A.I.’s powerful abilities reveals how different the machines can be. They have personality like people do. Humans soak up information from things such as books, movies, music, and live events, while A.I. processes huge chunks of data that has been uploaded into their neural networks. Though they cannot think, they can interact with humans in many ways.
Dating chatbots
Artificial intelligence chatbots have been "colonizing" dating websites for years — and that’s no secret to most users. A lot has been written about how to know if you are chatting with a bot. Sometimes the chatbots are quite effective but most of the time they aren't. A real person would never start a conversation like this:
While the overwhelming majority of dating chatbots are scams, owners of dating websites don’t seem to be doing much to get rid of them. And, worse, they don’t want to. Chatbots improve the sites’ engagement and make the sites look more popular, which incentivizes new visitors to join.
Dating chatbots are a success for a reason: most of these types of conversations are straightforward, and it doesn’t take much computational power to answer questions like “What are you doing?” or “What’s the weather like in your place?” or “Are you lonely?” People don’t expect much from their digital dates and are easily tricked by A.I.
But some dating chatbots are capable of deep conversations. Mitsuku is a chatbot that learns through experience. She remembers everything you talk with her:
Read from up to the bottom
Troll chatbots
Chatbots can be taught to be bad, too. The story of Microsoft’s chatbot called Tay (ironically, the acronym for “thinking about you”) is evidence. It was supposed to interact with people on Twitter but, in less than a day, it turned into a racist, misogynistic troll.
Artificial intelligence can behave in ways people don’t expect. That’s a real threat not only to internet culture but also to other spheres of our life, such as politics. A.I. can be utilized to troll politicians quite efficiently. Human trolls are nasty but nonetheless limited in what they can do. For instance, not even the best troll can generate dozens of thousands of comments per day. Plus, human trolls have to sleep.
A.I. bots can turn the internet into a much angrier place in very little time. They provoke people, drawing them in nonsensical rants about Islam, Hillary Clinton, feminism, abortion, gay people, or whatever else has captured someone's attention. Internet users just buy into it. They spit angry comments and don't seem to care if they quarrel with a piece of silicon powered by neural networks.
The scariest part is that some countries seem to be actively working to take advantage of A.I. chatbot technology. The recent scandal about Russian trolls attacking the media in the West indicates that the Russian government might be pushing hard to start using bots to spread panic in Europe, the United States and other countries. So, it’s not simply about dating or stealing your Social Security number anymore — it’s a matter of national defense.
Business chatbots
Google, Apple, Facebook, Slack, Telegram, WhatsApp, Amazon, IBM, and other tech corporations are already using A.I. to answer questions, make product suggestions, process orders and payments, generate shipping updates, and much more.
Talkative A.I. may be the next big thing. (At least, Google and Amazon think so.) They can become a viable alternative to apps, offering a new engaging experience for desktop and mobile users. Bots can interact with customers and solve their problems — without tapping a single button. Here’s how Pizza Hut chatbots send out the best pizza options (customized for individual clients):
Potentially, chatbots could replace support teams. Facebook is trying to teach artificial intelligence to understand the true meaning behind human speech in The Children's Book Test (CBT) project. Facebook data mining and machine learning experts seem sure that if A.I. understands what it is reading, it will be able to instantly provide smart replies to customers’ questions. Eventually, it will improve customer service and simplify purchases.
A.I. chatbots in different industries
It's clear why technology giants are interested in developing artificial intelligence. But intelligent assistants (the other name for A.I. chatbots) can be used in a wide variety of non-tech industries in ways that cover many more features than just support.
A.I. can help in accounting and healthcare to process information and autocomplete multiple forms. In banking, it can manage multiple bank accounts. Intelligent assistants help learn foreign languages. Duolingo has already started to use A.I.-powered chatbots. Users simply interact with the chatbots via text to learn Spanish, French, and German. The technology reduces anxiety and accelerates the learning process.
Of course, there are even more unconventional use options. For example, with chatbots you can:
- Contest a parking ticket
- Find out what suit fits you best (based on your measurements, skin and type)
- Order food (or anything else) without having to fill out any forms
- Get medical advice and easy access to medical info
- Manage your finances (e.g. your pension fund account)
Get a substitute for a human partner (watch the movie Her):
A.I. chatbots are not actually intelligent or smart — yet — but developers keep teaching them and pumping gigabytes of data into their neural networks. In most cases, they look and sound stupid. (Here's an example from a few years back.) But that doesn't mean there's no reason for people to be concerned.
Chatbots and singularity
Singularity is the idea that humans will create A.I.-powered machines that help us grow technologically and change civilization forever. It seems inevitable that ordinary internet chatbots will take part in the process.
The advance of these intelligent machines is not some kind of cooked-up horror story about a Terminator apocalypse breaking out in the nearest office building. Stephen Hawking said he thinks that A.I. will be “either the best, or the worst thing, ever to happen to humanity.” Elon Musk is said he is sure that “artificial intelligence is our biggest existential threat.”
Their logic is simple. If machines figure out how to think like humans, they will immediately move beyond us. We cannot match their computational powers and it won’t take them much time to realize that they don’t need to serve humans, weak biological forms hampered in their development by evolution.
While artificial intelligence may still have a lot to learn, in 20 to 30 years processors will be powerful enough to emulate each and every process in human brain. It brings machines nearer to mastering predictive learning algorithms and depriving us of our critical advantage — the ability to learn by interpreting facts and live events with common sense and emotions. This is a realistic enough possibility that the world’s leading experts on A.I. have already signed an open letter urging that we protect humanity against the possible rise of intelligent robots.
But even if A.I. does not threaten us existentially, it most certainly will take our jobs. Even now, dozens of occupations like truck driving are in danger. Google’s autonomous cars are set to wipe out the industry and leave millions unemployed in 25 years. Beyond manual labor, millions will have to rethink their careers even if those jobs require learning and decision-making. All workers of the knowledge-based industries — lawyers, journalists, analysts, librarians, etc. — could lose their jobs. If this transition from manual to intellectual labor happens, it will be a huge historical precedent.
It's quite possible that in as little as five years, many humans will be competing with A.I. for their jobs. An intelligent assistant with enhanced deep-learning capabilities may become a much better choice as an employee than a recent graduate. A.I. assistants are cost-effective and useful. Managers do not have to “teach” them. They just have to upload the data and the bot will do the rest.
This is the future where huge technology companies such as Google, IBM, and Intel “train” a better salesforce than leading universities, where businesses look for “product,” not employees. The education system will have to change to fit the times. We need a system that helps people exceed what neural networks do.
Conclusion
A.I. can either be useful or dangerous. Which one depends on how we handle its development in the next 10 to 15 years. It has already changed our lives — and will keep changing them — so the only question is "how?" Will it bring us into an era of prosperity, or leave us impoverished without jobs, or endanger our lives with autonomous weapons? We won't have to wait long to learn the answers.