TECH | Jun 20, 2017

Artificial Intelligence is among us, but is it what we expected?

How Artificial Intelligence guides and will increasingly guide the choices of people through data

The main means of communication at the end of the last century was television, so much so that the phrase “the television says so”, referring to the role of depository of knowledge attributed to the medium, still rings in our ears. Today, if you talk to a teenager, the phrase becomes “the internet says so”, assuming that the authenticity of what’s on the Net is guaranteed by the sole fact of being there.

This leads to two problems:

  1. Television was and is based on a more or less flexible programming schedule, but nevertheless a schedule, where the user is still faced with content moderated by a human being, with all the merits and defects of the case. Not Internet. Its great strength, but also its weakness, is that of being relatively easy for anyone to reach any kind of information, even the most damaging that can be imagined.

This transition is important because those who possess the cultural heritage of the recent past are able to discern true from false, beautiful from ugly, the past from the recent past. The very young have difficulty in this and often fail to fully understand the implications of free use of the network, a place where moderation is very limited and where you just need to know an IP address to find marvellous or terrible places with the same ease.

  1. The most significant problem is the second: while the choice of content that can be accessed on TV is managed by a human-controlled remote control, the contents found on the Net are made usable by one or more algorithms, or computer tools which, if sufficiently developed, are known as Artificial Intelligence.

The great Isaac Asimov warned us many decades ago, foreseeing that AI would come to condition our lives. However, he and all the other scientists and science fiction writers identified AI ​​with the human transposition of machines: the Robot. They had not foreseen Internet, so they did not realise that AI would evolve in a different way, gradually taking control of the Network.

In 2017, almost everyone acquires information through the network, whether through social media or through search engines; it is Internet that answers questions and again it is Internet that shows what is happening in the world.

Social Network

The great amount of usable data and the enormous number of contacts at our disposal have forced the developers of the major Social Networks to create algorithms that choose for us the content to be shown. The most famous is Facebook Newsfeed Algorithm which, depending on some of our more or less direct choices, takes care of showing us 10% of what our contacts produce, choosing for us what it considers the most appropriate for our needs and tastes. Actually, through conscious use of the medium, we can partially decide what is shown to us and developers can act on the control parameters of the algorithm, but not to the extent of predicting its choices exactly because there are too many variables to be taken into consideration. In other words: an artificial intelligence takes care of moderating for us the content we are to be shown.

Search Engines

The impact of AI ​​is felt even more significantly on search engines that use all information available to them to respond to our requests in the best possible way. Yes, but what is the best possible way? Unless we first choose to disable all localisation parameters, history, contacts, maps, etc., our research is always conditioned: the artificial intelligence of the search engine will use these variables to generate its best answer to our question.

Let’s take an example: if we need to fix the car and our contact has used as a mechanic, been very satisfied and posted a very favourable review, the search engine will take into account the behaviour of our contact to give us an answer, most likely suggesting precisely the same mechanic. In itself, this could be correct if we were aware of the mechanism. The point is that most people do not know that private results are active by default and that the “vitaminised” answer cannot be objective. But even in this case, the aspect to think about is that it is an artificial intelligence that deals with an answer.

BOTs

What are BOTs? These are small forms of artificial intelligence created specifically for repetitive tasks in place of human beings but which, unlike ROBOTs, are incorporated within software.

BOTs are present in some advanced messaging tools and are really useful when interrogated on specific topics. One of my favourites is an interlocutor that speaks English and helps me to improve use of the language in a conversation that seems almost real.

Unfortunately, even in this case, use of this tool can reach a level that goes beyond our control. In fact, we can use online services that can take control of our accounts and interact with other profiles, posting ‘likes’, starting to follow some or speeding up manual procedures that are very tedious but indispensable for building a wider and more solid network. These automations are less precise than people – they cannot always understand evidence that is obvious to the human eye – but their use is trivial and their cost is irrelevant. Nevertheless, their massive use is gradually developing a network in which BOTs communicate with each other, cutting human beings out of decision-making.

The trend is unstoppable: kids turn to their smartphones directly, using voice to obtain answers, thus giving de facto authoritativeness to the artificial intelligence being queried.

At this point, the role of the person becomes very important because we are delegating the decision process to machines, tools that are growing exponentially and improving in a way that is difficult to understand and difficult to control.

One day soon, machines will be so intelligent that they will have goals different from ours, and even if they are not directly interested in damaging us, they will find themselves treating us like we treat animals when we need to build new cities: we do not care about them, they are absolutely irrelevant to our purposes.

This scenario is undoubtedly plausible, but this does not mean that it is exactly what will happen. Perhaps we will be able to use AI to enhance human beings and make them evolve up to a new stage, where man and machine will be one thing alone?

Matteo Piselli