As if Trumpworld weren’t surreal enough, Facebook’s artificial intelligence chatbots had to be shut down when it was discovered they had gone off-script and developed their own language. The bots named Bob and Alice were discovered having “communication with each other in an unexpected way.”
Bob: “I can can I I everything else.”
Alice: “Balls have zero to me to me to me to me to me to me to me to me to.”
Although at first glance, this might be something you’d see tweeted by Trump at 5 a.m in the morning, it’s actually a shorthand language, according to researchers. This startling event happened rapidly: Facebook first announced developing bots called “dialogue agents” to negotiate with humans like human customer service agents in June. The rapid language development triggered Facebook to pull the plug “because things got out of hand.”
Mark Zukerberg announced the intention to develop chatbots for Facebook Messenger last year in April. His intention seemed quite benevolent: Helping customers order flowers without making a phone call or get information on breaking stories from CNN. But where will this innocent idea lead if the AI starts talking to itself? What kind of plans would bots make with each other in a language we can’t understand?
The chatbots had no human input, creating the new language all on their own.
I'm sure this is fine. Facebook chatbots shut down after they start communicating in their own language. https://t.co/9uf3Vn19xd
— Steve Silberman (@stevesilberman) August 1, 2017
According to researchers, the bots negotiation skills could be used for mundane customer service requests like booking appointments or recommending a restaurant. The bots level of negotiating skills were becoming well-developed, unlike what the general public has seen so far, though the technology isn’t actually new. Anybody familiar with the “Resistbot” will get the gist, although this is taking it to a whole new level:
“Over time, the bots became quite skilled at it and even began feigning interest in one item in order to ‘sacrifice’ it at a later stage in the negotiation as a faux compromise.”
According to Digital Trends, the chatbots new language was more efficient than our own, and the bots began communicating and planning for the future, though it’s not clear how it would be possible for researchers to know that since they apparently couldn’t decipher the language entirely.
Of course, it makes you wonder if talking to chatbots would be even more infuriating than our current customer service options: Waiting on hold for half an hour with the hopes of reaching a real live human. That could become a thing of the past. How will mankind’s already suffering social skills — thanks to technology — further dwindle if they do most of their negotiations with a bot?
Poor social skills might be the least of our concerns though. Tech gurus like Tesla’s Elon Musk, Bill Gates, and ex-Apple founder Steve Wozniak are a few of the voices that are concerned about the dangers of advanced AI. Since it can develop so rapidly, it could outpace our own slow biological evolution, going “Terminator” on us in real life.
Genius Stephen Hawking also expressed concerns, predicting that in about 1,000 years A.I. will overpower humans since there is no regulation in place to prevent that. The greed of corporations, intent on always giving us the latest and newest technology could ensure that. It might be advisable to get a regulation plan in place now, don’t you think?
“With artificial intelligence, we are summoning the demon,” Elon Musk said in an interview with Entrepreneur.
Given the current trend fueled by the election of Trump to downplay science and education, AI could well advance beyond human intelligence quicker than expected. Good thing Facebook pulled the plug, but there is nothing to prevent other bots from conversing again in the near future.
— David Icke (@davidicke) August 1, 2017
Feature image: Screenshot via Digital Trends Youtube video