The chatbot dilemma that isn’t a dilemma

The presentation of Google’s experimental Duplex robotic assistant technology at the opening conference of Google I/O 2018 has already made headlines. Immediately after the demonstration of two calls to make reservations, one at a hairdresser and another in a restaurant, carried out by an assistant with two different voices that were impossible to identify as non-human, critics have described the company as “insensitive” and its innovation, “uncanny” or even “horrifying (…) ethically lost, rudderless and has not learned a thing.”

Such responses reflect an inability to think the question through: why assume that having to talk to a machine unawares is some kind of insult, when in reality we’re dealing with a commercial transaction and that as soon as the service industry is able, it will start using chatbots. Have businesses ever had any kind of problem in deceiving their customers and making them think they are dealing with a human when they are not? The litany of clumsy attempts includes automated robocalls to mail merges and beyond. If companies can save costs by passing off a robot as human, they will. In business management, the costs justify the end.

Indeed, what worries me most about Google Duplex is how it will be used by businesses: telemarketing by irresponsible companies (the perfect employee who does not mind making the same call a hundred times, has no fear of rejection and any number of sales pitches devised by machine learning from millions of calls) and the evidence that there is always a company with a manager stupid enough to do what no one with a minimum of common sense or intelligence would do. Instead of which, the first response from a few well-meaning individuals is to worry about the ethics of a hairdresser or waiter not knowing if she’s talking to a potential client or her robotic assistant …

Given the criticism and requests to make it mandatory for robots to identify themselves as such in their conversations, Google says it will incorporate a disclosure along the lines of an initial warning making it clear we’re talking to a robotic assistant. Who knows, maybe we’ll have to pass a captcha or a Turing test to prove we’re human before making a call, which will be used as an excuse when anything goes wrong (we’ve been blaming computers for human error for decades now).

In other words, we create a technology to make life easier but instead of embracing it, we now worry about the rights of the human being who picks up the phone and doesn’t know whether he’s talking to another human or a machine …

I repeat: what businesses want to do is SELL THINGS, and as quickly and efficiently as possible. Dealing with an assistant is more efficient and it will make fewer mistakes than we humans, cause fewer problems and so the transaction will take place more quickly. There is no ethical dilemma about whether the employee knows if she is talking to a machine or not, and pretty soon, the company will replace that employee with a robotic assistant. In the meantime, nobody is going to come to any harm by talking to a robotic assistant.

Let’s not create problems that don’t exist. The only interesting thing about all this, if you’re a sociologist, is perhaps how our behavior and our protocols will evolve in a world in which the voice of machines is indiscernible from that of people, or whether we will overuse our robotic assistants, asking them to do things no human would naturally consider.

Some of the reactions to this latest development have been just plain stupid. Let’s develop technology without constraints, let’s see what problems emerge from its use and then try to correct them by introducing improvements gradually through trial and error. This is how progress works, not jumping to conclusions about the immorality of having to talk to a machine. If you don’t like the sound of Google’s robotic assistant, don’t use it. But leave the rest of us in peace and do not pose absurd moral dilemmas that don’t exist.

(En espa?ol, aquí)

?