Claudia Borg
We are in continuous contact with Artificial Intelligence – AI. We often hear that everything we touch is AI – our phones, voice assistants, Google Translate, the playlist on YouTube or Spotify, Facebook’s newsfeed, weather forecast. Yet, can we really say that the machine is actually thinking in order to provide us with these services?
Recent advances in AI, through the use of deep learning and improved computational power, make it seem like we are interacting with intelligent systems that are able to predict what we like, want and need. Current AI systems are based on algorithms that have learnt to do really specific things by being trained on thousands or millions of correct examples (e.g. identification of objects in an image).
Some systems have learnt a specific task on their own, without the examples, but simply by being told whether an outcome satisfies certain criteria (e.g. in game AI). And in other instances, we simply leave the AI to learn the patterns on its own (e.g. the modelling of a language). In some specific tasks, AI systems can be even more accurate than humans.
These advances force us to take a step back and ask whether we are really interacting with a thinking machine. After all, this has been the Holy Grail of AI since the beginning – to produce a machine that can think and do things like a human can. And while the same AI system can be set to train itself on different tasks (e.g. Alpha Zero can play Go, Shogi and chess), we still cannot have a single system that can learn and do all tasks. So there’s no comparison to a human brain – a single ‘system’ that can learn many different things.
In 1950, Alan Turing set out what he considered to be the signs of a thinking machine, one that we would call intelligent. He proposed the ‘Imitation Game’ and proposed that if a machine manages to pass as human, then it is said to have intelligence. For Turing, the measurable thing about intelligence is the output, rather than the process itself that derived that output.
However, two years before Turing’s test, Geoffrey Jefferson, a pioneering brain surgeon, set the bar higher than Turing. He stated that a machine cannot be said to have intelligence by the chance fall of symbols, but rather by producing something through the process of thoughts and emotions, and therefore not only produce something, but also know that it had produced it.
We still have not arrived at a conclusion as to how to exactly define whether a computer system is intelligent or not. This aspect raises many controversial discussions and disagreements. However, it is very clear that despite the advances made in recent years, there is still a long road ahead in what can only be described as an exciting and fast-evolving field of research.
Dr Claudia Borg is a lecturer with the Department of Artificial Intelligence at the University of Malta. She is the lead coordinator for LT-BRIDGE, a H2020 project aimed at integrating Malta into European Research and Innovation efforts for AI-based language technologies. For more information visit https://lt-bridge.eu/
Did you know?
• Earth is the only planet not named after a Roman god or goddess, but rather associated with the Greek goddess Gaea (Terra Mater).
• The name Earth comes from Old English and Germanic, derivedfrom ‘eor(th)e’ and ‘ertha’ which mean ‘ground’.
• The Earth’s moon might seem dead and inactive. But in fact, moonquakes, or ‘earthquakes’ on the moon, do occur.
• Although less common and less intense than those on Earth, scientists think that moonquakes are related to tidal stresses associated with the varying distance between the Earth and the moon.
• The longest mountain range found on Earth is underwater, called the mid-ocean ridge. It is a chain of volcanoes that spans 65,000 km and rises 18,000 feet above the bottom of the sea.
For more trivia see: www.um.edu.mt/think
Sound bites
•Scientists have analysed data in an effort to understand how the internal state of the body influences the brain’s decision-making processes. They found that two of the brain’s decision-making centres contain neurons that may exclusively monitor the body’s internal dynamics. This suggests that the brain’s decision-making neurons are wired to constantly monitor and integrate what is happening in the body, explaining why, when we are anxious or stressed, we are more likely to make bad decisions.
https://www.sciencedaily.com/releases/2021/08/210830140224.htm
• Climate change is not only a human problem; animals have to adapt to it as well. Some animals are shapeshifting and getting larger beaks, legs and ears to better regulate their body temperatures as the planet gets hotter. However, animals are facing unusual pressure since these changes are happening over a much shorter timescale.
https://www.sciencedaily.com/releases/2021/09/210907110718.htm
For more soundbites listen to Radio Mocha www.fb.com/RadioMochaMalta/