Artificial intelligence – 5 terms to boost your real intelligence

Artificial intelligence, or AI for short, has been all the rage in recent times. A few years ago, self-driving cars showed great promise and there were predictions that automated cars would become the norm very soon and taxi drivers would lose their jobs.

Predictions about that particular branch of AI proved too optimistic (or pessimistic, depending on your point of view). But that hasn’t dampened the enthusiasm about AI as AI-generated text and pictures are the new kids on the block, astounding everyone with their ability to produce art and write essays that could pass off as the work of a real-life person.

While this article won’t make you an expert in the technology, it’ll introduce some relevant terms so that you can follow along during conversations. Perhaps you’ll even get good enough to drop in a few phrases at the right moments. We’ll even throw in some examples of AI right here in Singapore!

1. Machine learning

As its name suggests, machine learning is all about enabling a machine to learn and is what is driving the gigantic improvements in AI in recent times.

Instead of learning by rules that are explicitly programmed, machine learning allows a computer to learn through examples. Imagine having to teach a child what a cat is. You could spend all day describing how a cat has four legs, a tail, whiskers, etc, and chances are the child might not even recognise a cat when he or she sees one in real life. Or you could show the child five pictures of a cat and voila, he or she will be pointing out cats at your HDB void deck.

While it takes thousands of pictures for a machine to learn to recognise a cat, it turns out that if you show enough pictures of a cat to a machine, it eventually is able to process a picture of a cat it’s never seen before and correctly say, “Aha! That’s a cat!” The general principle is what’s driving the AI behind human detection in surveillance cameras, automated transcribing of speech, and spotting signs of diseases in medical scans.

The explosion in data production means that machines have ever more data to consume, and therefore more examples to learn from.

2. Neural networks and deep learning

Neural networks are a more complex form of machine learning that uses nodes to process data. These nodes are interconnected and form layers that, taken together, form the neural network. Data is processed within the nodes in a layer and then passed on to the next layer until output is eventually generated from the final layer.

So a neural network that is fed pictures of cats would ingest each image through its layers, spotting patterns among the many pictures and enabling it to learn that cat pictures usually have certain features.

If you’ve heard the term deep learning, it simply refers to a neural network with more than three layers. Some deep learning models are so complex that they are described as black boxes, and even the people who created them cannot fully explain how they work.

3. Generative AI

AI has gone from absorbing content and making predictions to creating the content itself. That’s essentially what generative AI is.

Need a hyperreal graphic of a parrot made of carrots? DALL-E can create it with just a few text prompts. Feel like reading a summarised version of a literary classic with some analysis of its themes? ChatGPT has your back.

One of the amazing things about generative AI is its ability to understand prompts that are written in plain English, just like the way you would make a request to another human. And that’s largely possible because of the next topic.

4. Natural language processing

This branch of AI is all about enabling computers to understand text and speech the same way humans are able to.

The main challenge of NLP is in determining the intended meaning when confronted with ambiguity, sarcasm (“I’m sooooo impressed”), idioms (“He’s over the moon”), and metaphors (“An ego too big to fit this room”).

Again, deep learning is powering the improvements in NLP, enabling machines to consume voluminous writing and audio clips to enhance their ability to predict what word should come next.

Think how much better it is to pick up Thai by immersing oneself in Bangkok for months instead of spending the same time reading a grammar textbook and a dictionary, and you get the idea.

NLP is what’s behind tools like Google translate, Siri and Alexa, and chatbots.

5. VICA – Virtual Intelligent Chat Assistant

If you have visited government websites, you would probably remember Ask Jamie – GovTech’s virtual assistant, which was rolled out in 2014. That said, her days are numbered.

Jamie’s replacement, VICA, is already out answering citizens’ queries in the wild. Armed with next-generation NLP, it’s already powering the chatbots on multiple government websites.

At the time of publication, VICA was in operation for IRAS (Ask IRAS), EMA (Ask EMMA), and ICA (Ask ICA).

There are also plans for Singpass integration, live chat escalation, and support for multiple chat platforms like Whatsapp and Telegram.

Time for some human learning

AI's encroachment on creative, white-collar jobs has left many professionals feeling uneasy about their future.

However, in the face of uncertainty, we can equip ourselves with knowledge and face this challenge head-on. Let's embrace the ever-evolving tech landscape, seizing the opportunities it presents.

By understanding AI's intricacies and seeking ways to collaborate, we can harness its power to amplify our creativity. Stay positive and curious, never ceasing to learn and explore. Together, we can shape a future where human and artificial intelligence harmoniously coexist, opening doors to remarkable achievements.

As Alan Turing wisely said, "We can only see a short distance ahead, but we can see plenty there that needs to be done."

Let's embark on this transformative journey with resilience and optimism.