(Bloomberg) -- Got bot? The news sometimes seems full of bots, often in alarming ways, from the bot-powered Russian trolls that invaded social media during the 2016 U.S. election to the software bug that briefly had Alexa, the voice-based virtual assistant of Amazon.com Inc.’s Echo home speakers, breaking out in random laughter.
1. What’s a bot?
It’s a computer program that executes automated tasks over the internet. Think of it as software’s version of the steel-and-cable robots that weld cars or assist surgeons.
2. What kind of tasks?
A simple bot might tweet a weather report once an hour or, somewhat less innocently, search for and sweep up tickets to popular concerts for resale at higher prices. Many of the bots we encounter these days are so-called chatbots, which are programmed to mimic human dialogue. On Facebook, WeChat, Twitter or a wide range of customer-service interfaces, chatbots can perform relatively simple tasks like placing food orders for users or booking airline tickets. Amazon’s Alexa, Apple Inc.’s Siri, Microsoft Corp.’s Cortana and Google Assistant are examples of chatbots.
3. Why are people worried about them?
Nefarious bots have been around for some time, posting spam or inflating traffic to websites, distorting product rankings on e-commerce sites or writing false reviews. Recent attention has turned to how bots might sway public opinion in campaigns or on other weighty issues. Bots can be programmed to use social media to post on a massive scale. Tied together into "botnets” of tens of thousands of accounts, these non-human actors can work in concert to make social-media posts go viral.
4. Is that what happened during the 2016 campaign?
In part, yes. Twitter Inc. found that Russian-linked bots shared candidate Donald Trump’s tweets 470,000 times during the final months of the 2016 election (versus 50,000 shares of tweets by his opponent, Hillary Clinton). More broadly, Twitter said its review identified more than 2 million automated, election-related tweets from Russian-linked accounts, which collectively received about 455 million impressions within the first seven days of posting.
5. What can be done?
Companies have been trying to clamp down on such activity, but it’s not easy for a website to distinguish bots from humans -- especially since some bot activity is intentionally piggy-backed on human-created posts. Many websites use some version of the human-verification test known as Captcha, which poses a task -- like solving a puzzle, or retyping often-blurry words -- that’s supposed to be easy for humans but a lot harder for robots. But technological advances are making the bots more and more capable of defeating such safeguards.
6. So what’s next?
Artificial intelligence is making bots more powerful and more lifelike, for better or worse. In a May demo, Google had Google Duplex -- an extension of its Google Assistant chatbot -- call a hair salon to book a session for a client. It carried on a conversation with an employee who appeared unaware who, or what, she was talking to. At one point, the chatbot responded to a request to hold by saying, "Mmm-hmm." Reactions were divided between those wowed by how lifelike the bot sounded and those horrified by the same thing. Google later said the bot will identify itself and tell people that the call is being recorded, at least in locations where that’s required.
The Reference Shelf
- QuickTake explainers on cyberwarfare and online deception in the 2016 U.S. campaign.
- Twitter bots are roiling the Middle East, according to a Washington Post column.
- Amazon says Alexa won’t laugh as easily anymore.
- A "bot army" may have bought 30,000 tickets to "Hamilton" on Broadway.
- A Gizmodo video on bots.
©2018 Bloomberg L.P.