In 1950, Alan Turing posed the question: will there ever be a machine that can successfully imitate a human? In recent years, several machines have come close to passing what is known as the Turing test. Now a patent filed by Google has revealed that it is working on a machine capable of not only imitating humans, but of imitating particular humans.
The big idea is to create software that can analyze an individual’s writing style across a series of social media messages and then go on to create new messages in the same style. This would mean it could post on Facebook, Twitter or similar on the person’s behalf; if it did it well enough, those reading the message would not know the difference.
This may sound a little creepy where it applies to personal relationships – should we really be depending on computers to keep up our friendships on our behalf? When it comes to social media marketing, however, it could be incredibly useful, enabling businesses to keep up with their customers much more effectively. People who have managed social media accounts for long periods of time will realize that many of the responses appropriate for customer comments are essentially the same. Why not have the computer take care of them? If it can’t be detected, the business will get the benefit of customers feeling that they have received personal responses, leading to them feeling valued. It could save a great deal of time and money in comparison to having staff members do the job.
There are risks, of course. Computers are always likely to struggle with things such as sarcasm, and giving a response that gets it wrong or seems too flippant could potentially cause a lot of trouble. Since the software can be expected to improve over time, the real question is when companies should take a chance on the benefits outweighing the risks, and how much autonomy to give it. There is a possibility that there will be the option of having it make suggestions rather than posts, which will still save time if vetting is quicker than writing.
How to balance these matters will depend on the nature of the business involved. The software will be most advantageous where small companies with few staff are serving a large customer base. Businesses that rely on repeat business or on having a good reputation in a niche market would be advised to be cautious about first-generation products.