Artificial intelligence is a burgeoning field that is finding its way into all aspects of modern life including self-driving cars, smart assistants, healthcare, disease tracking, automatic financial investments, booking travel, counselling therapy, monitoring social media, and much more. It’s a powerful, fascinating technology that is still in its infancy, but it has already demonstrated its potential to profoundly transform the way we work, play, and live.
How does it work?
AI systems are a bit like sophisticated and very well-informed guessing machines. Take chatbots for instance. They can successfully mimic the kind of exchanges that you might have with a human virtual assistant because they have been fed huge amounts of data and they have learned from it. This training data is all labeled, so the AI system is able to analyze it for patterns and correlations. Any patterns that it does find can help it to predict the most applicable responses. The more quality data they are fed, the better their outcomes, whether that’s for text, sound, or image recognition.
The trio of core cognitive skills that AI programming focuses on are learning, reasoning, and self-correction:
This stage is concerned with the acquisition of data and rule creation. The system learns how to take data and transform it into information that can be actioned. Algorithms will be a familiar term to most people, but not many of them will be able to offer as succinct a description as this—that an algorithm is a set of rules that machines use to perform a particular job.
Achieving the right result comes down to programming the AI with the correct algorithm.
This is about fine-tuning algorithms. By constantly tweaking them for improvements, it’s possible to reduce errors and improve performance over time. The more it refines, the better it gets.
What are the pros and cons of artificial intelligence?
Artificial neural networks and deep learning artificial intelligence technologies are evolving at a fast pace because they are so good at what they do. As clever as we humans are, we can’t match machines for their ability to analyze vast datasets and use them to predict outcomes.
AI apps that use machine learning can churn through more information than a human could ever get through in their entire lifetime, turning it into workable information without breaking a sweat. The only downside at the moment is that this supreme efficiency doesn’t come cheap, but that will no doubt change in the future.
- Well suited to detail-heavy jobs;
- Handles data-heavy tasks faster;
- Offers consistent results;
- AI-powered chatbots never rest.
- Needs extensive technical understanding;
- AI experts are in short supply;
- System can only work with what you tell it;
- Inability to generalize between tasks.
Strong vs. weak AI
AI is described as being either weak or strong.
- Weak (a.k.a. ‘narrow’) AI is built to do just one job. Systems like commercial robots and chatbots use weak AI.
- Strong AI, (a.k.a. artificial general intelligence or AGI) is designed to have more in common with human cognitive functions. Humans are great at applying learning from one area to another, and in the same way, a strong AI system leverages fuzzy logic to work out solutions on its own. Theoretically, a strong AI program should be able to trick its way through a Turing Test and the Chinese room test.
The four kinds of AI
- Type 1: Reactive machines. These AIs make decisions but don’t remember what they’ve done before, so we can’t learn from them. They’re good at handling one task, like the chess program Deep Blue, which is good at recognizing chess pieces and predicting moves, but it doesn’t remember or learn from its previous ones, so it can’t use that knowledge to help it make better moves in future.
- Type 2: Limited memory. As the name suggests, These AI systems do remember things to an extent, so they can use what they’ve done before to help shape what they do next. The way that self-driving cars make some decisions is down to this kind of limited memory.
- Type 3: Theory of mind.In an AI context, this psychological term is describing a system that is socially intelligent enough to comprehend emotions. It will be able to gain inference of humans’ intentions and predict their behavior, paving the way for them to fully participate with humans.
- Type 4: Self-awareness. This may never come to pass, but it’s the idea that an AI will be able to utter Descartes’ famous dictum, “I think therefore I am” (cogito, ergo sum) and mean it!