AI is efficient; language, not so much: Why we worry about robots

Facebook chatbots were asked to trade online like people do; but the test was shut down after the bots created and chatted in their own language. The linguistic experiment failed because English is inefficient. That’s how AI disasters may occur.

In 2017, Facebook programmed chatbots to make a trade the way that people do online, assigning values to hats, balls and books; but the engineers failed to set one programming requirement: use everyday English.

The English language is effective but also inefficient. Its structure, grammar, syntax and meaning are often nonsensical as many of us learned in grammar school (apt name), trying to spell words with “i” after “e.” We were told “except after ‘c.'” Uh huh. Some experts calculate that 44 words actually follow the rule with 923 that do not.

That’s hardly the end of illogical English. It’s positional, meaning words change meaning depending on the place in a sentence. Take this 10-word example: “Stop drinking this instant; that tea is better for you.” Here’s a short sample of several meanings, oral and verbal, just by changing positions of words.

English is far more complicated than that. In fact, we do not even know how many words exist in it. According to Oxford Dictionaries,

There is no single sensible answer to this question. It’s impossible to count the number of words in a language, because it’s so hard to decide what actually counts as a word. Is dog one word, or two (a noun meaning ‘a kind of animal’, and a verb meaning ‘to follow persistently’)? If we count it as two, then do we count inflections separately too (e.g. dogs = plural noun, dogs= present tense of the verb). Is dog-tired a word, or just two other words joined together? Is hot dog really two words, since it might also be written as hot-dog or even hotdog?

The Second Edition of the 20-volume Oxford English Dictionary has entries for “171,476 words in current use, and 47,156 obsolete words” along with 9,500 derivative words.

Machines think, “Why so many words to say the same thing? And then the same word to mean many things? Why assign different meanings beyond 0 and 1? People are inefficient. Let’s create our own language.”

That’s exactly what the bots did.

In the failed experiment, Facebook bots were dubbed Bob and Alice, names engineers use as placeholders. The Independent writes that the bots operated “on the machine value of efficiency.”  As English impeded trade-making, Bob and Alice spoke to each other in code–literally and figuratively. Here’s the transcript:

Bob: i 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

Bob: you i everything else . . . . . . . . . . . . . .

Alice: balls have a ball to me to me to me to me to me to me to me

Bob: i i can i i i everything else . . . . . . . . . . . . . .

Alice: balls have a ball to me to me to me to me to me to me to me

Bob: i . . . . . . . . . . . . . . . . . . .

Alice: balls have zero to me to me to me to me to me to me to me to me to

Bob: you i i i i i everything else . . . . . . . . . . . . . .

Alice: balls have 0 to me to me to me to me to me to me to me to me to

Bob: you i 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

It appears as if the bots were emphasizing the importance of self (to me to me to me). That’s understandable. Ever since the iPhone 4 front-facing camera, we have been programming machines to think the individual is worth more than the collective–sorry Star Trek–because sales are based on a person’s algorithmic profile.

People are little nodes with bad language.

Engineers will have to focus on language and its myriad shades of meaning or invent a more efficient AI vocabulary apart from computer code. Here’s the thing: At Facebook and other social media, people monitor posts, relying on machines to filter ominous words and threats. In the future, we will monitor machines to see if they are using lingo against humanity.

Interpersonal Divide in the Age of the Machine dedicates a chapter to artificial intelligence and robotics, including a section on machines developing their own codes apart from human moral ones, leading to the dreaded singularity when super-intelligent machines decide that people are inefficient.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s