Human imagination is like a box with a wide range of ideas: every person could imagine different things, but as a society it is common to picture outstanding scenarios where the species would achieve amazing progress. For instance, some movies like X-Machina or Bicentennial Man envisioned a future where relationships between robots and people are possible, and implanted the idea of machines having consciousness and feelings towards others. We probably looked at these films as impossible scenarios, but what happens when reality overlaps with fiction?
A few days ago, many people around the world started talking about LaMDA, a Language Model for Dialogue Application created by Google in 2017, system that recently gained relevance because it is claimed to be sentient by one of the researchers of the technological company. Considering the digitalization progress, it is not that hard to believe that the machines could develop unexpected capabilities, but before judging or commenting, it is important to understand how artificial intelligence work.
The Evolution of AI
Artificial intelligence is software that mimics human responses under certain conditions, and according to Alan Turing, “the possibility of letting the machine alter its own instructions provides the mechanism for this” (Turing, Lecture to the London Mathematical Society, 20 February 1947). While Turing’s definition is one of many for artificial intelligence, it emphasizes its sometimes unbelievable capabilities to learn as humans do so that it can address problems it has never faced before. AI as Turning imagined it is not yet here, but its precursors are and can be applied to different fields: from decision-making in a self-driving car to answers provided by a chatbot. Specialists have improved algorithms and, undeniably, neuronal networks have changed the rules.
It is important to remember that the first discussions and discoveries of artificial intelligence were in the 40s when engineers discovered an algorithm to decipher hidden messages during the war. If we compare that with the current decade where big data has undoubtedly gained relevance, we can say that artificial intelligence took great advantage of the massive amount of data available and the capabilities of the computers, which allow the machines to respond logically and, in some cases, faster than humans (e.g., in a chess game)
Nowadays, the software can understand different words and their meaning together in a sentence, learn from previous answers to develop a conversation, and reply according to the receptor’s assumed preferences in a very smooth way. Regardless of how amazed we are of this capabilities, the greatest limitation machines have is that they cannot transfer their “knowledge” on one topic to another similar or related one. For instance, if a human knows how to play chess, then the person would be able to play checkers in a few hours, but if a machine is trained to play chess, this is the only game it will be capable to play, and if we want the machine to play checkers, we need to train the software for that.
Taking into account the model’s mimic of the human brain and coming back to the main point, is it possible for a machine to feel like a human? The answer is no. Feelings are associated with emotions that occur within the body, while the machines can sense the world and agents around them, and by doing so they can respond to the circumstances. To make it clearer, software or robots can express sadness or happiness, but they do not feel it in the way that we know and do. LaMDA is a language built using artificial intelligence, and it is capable of developing answers according to what it learned. Undoubtedly it is an amazing creation, and it is bringing us closer to the superintelligence that scientists are looking for, but it is still a software created by humans whose functionality is learning from the experience and improving on its own.
What do we know about LaMDA?
To sum up, LaMDA is capable of connecting millions of sentences, giving them a meaning, and talking about many topics. Artificial intelligence needs humans as humans need computers. The two complement one another, but the existence of this new kind of intelligence would not be possible if the human does not indicate to the algorithm what to do and how to work. Although there is outstanding progress in this industry, we still have to keep the expectations of machines to have emotions on hold and leave this topic for the science fiction movies.
Read this article to know more about AI and its functionalities: How AI Is Changing the Playing Field With Predictive Analytics