Artificial Intelligence: What’s all the fuss about?

 

Article by The Minister of Truth

As a business, you could be forgiven for getting yourself into a right tizz about artificial intelligence (AI). It seems to be everywhere these days and everyone seems to be telling you that if you don’t harness its epoch-making abilities this very instant your business will be steamrollered faster than you can say ChatGP-what?!

 As with any new wave of disruptive technology, it’s all too easy to go into headless chicken mode and start making lots of panicky poor decisions as a result. So let’s all calm down a bit and take a deep breath.

 In this series of five features, we’re going to explain what AI actually is - the different flavours, the pros and cons, its capabilities and limitations, the opportunities and threats it presents. We’ll explore how certain iterations of AI could practically help different aspects of your business. And we’ll take a jargon- and myth-busting approach to a topic a friend once described to me as “sheer witchcraft”.

 

AI generated image of AI in the next 2 years

AI generated image of AI in 2030

AI generated image of AI in 2050

 

Let’s get one thing out of the way right at the start. Quite often people will say such-and-such service is “AI-powered” when what they really mean is that it uses machine learning. This basically means using considerable computational power to spot patterns in huge sets of data and then derive useful insights from those observations.

Think of those apps that help you identify birds, flowers and trees. They’ve all been trained on huge sets of visual and aural data looking for similarities in the mass of ones and noughts.

They’re still not 100% accurate – machine learning is still very clever guesswork – but they’ve got a hell of a lot better in recent years the more they’ve learned and refined their answers with the aid of human feedback. Translation apps are another case in point, even if they still sometimes struggle with idioms.

What’s caused the latest fuss is called Generative AI (GenAI), which makes use of Large Language Models and clever predictive algorithms to come up with – or generate - plausible-looking texts, images, videos and sound in response to our questions. You’re effectively tapping into the sum of human knowledge uploaded to the internet and generating ideas and images that can enhance your creativity and simplify your business processes. So potentially, such programs are incredibly powerful tools.

Some well-known programs include GPT-4, ChatGPT and DALL-E from OpenAI; GitHub’s Copilot and DeepMind’s AlphaCode, which help developers write computer code; Google’s Bard (still under development); Anthropic’ Claude; Synthesia; Grammarly Business; Jasper; Cohere Generate; and Copy.ai.

 

AI generated image

 
 

Exponential growth

Such is the dizzying speed of development in this area that there are now tens of thousands of generative AI tools on the market. Investment bank Goldman Sachs forecasts that investment in AI will approach $200 billion (£163 billion) by 2025 and that Generative AI could raise global gross domestic product (GDP) by $7 trillion (£5.7 trillion) over the next 10 years. Soon all business activities will incorporate AI in some shape or form. Can you imagine a life without Google search? Before long it will be impossible to imagine life without GenAI.

One significant limitation of ChatGPT was that, until recently, it could only access data up September 2021. It has now rectified this with a ‘Browse with Bing’ feature that enables real-time web browsing. It is also incorporating its text-to-image generator, DALL-E 3, into ChatGPT. Bear in mind that OpenAI launched ChatGPT – short for Chat Generative Pre-trained Transformer - in late November 2022. That’s how fast things are developing.

But what can it do?

So what can these tools actually do for us in the real world?

While scores of students are gleefully asking ChatGPT to write their dissertations for them, in business the opportunities to improve and personalise customer service; automate long-winded processes; and even predict what customers may want to buy or know about in future based on their previous behaviour and interactions, are exciting companies across the globe.

Chatbots have improved a lot in recent years but they can still be pretty clunky, often misunderstanding what we mean despite all the advances in natural language processing. But GenAI could make a step-change in this field. Imagine your own personalised chatbot that learns about you and understands you better each time, what your preferences are, your aspirations, the challenges you face. It could become proactive, offering advice, ideas, recommendations, reflecting the personality and values of your company. That’s a mouthwatering prospect. Think about the recommendation engines from streaming services like Netflix and Spotify, but much, much better.

We’ll go into much more detail about the practical applications of GenAI in future features, covering such topics as personalisation, omnichannel marketing, and security.

 

AI generated image of AI in the future

 
 

Abundance of caution

For now, it’s definitely worth standing back a little from the hype and adopting a testing, cautious approach. Generative AI is derivative by nature – it is only as good as all the data it has fed on. And some of this data is wrong, biased or just plain nasty. ChatGPT clearly warns users that sometimes it will get things wrong. These GenAIs programs even occasionally make up their own references in a phenomenon known as “hallucination”.

So before you dive in and start using conversational AI programs to enhance your customer chatbot experience, can you be sure it won’t go haywire and start spouting racist, sexist or just plain rude stuff to your customers? This is your company’s reputation on the line, and in the age of social media and instant outrage, any unfortunate gaffe from your AI-powered chatbot could result in a barrage of negative headlines and comments causing serious damage to your bottom line.

It’s important to understand that GenAI is still in its infancy, even if it is developing at an exponential rate.

Generative v General

We’re still some way from the rise of artificial general intelligence (AGI), however - sometimes known as strong or full AI. This is the version that eventually surpasses human intelligence, self-learns and develops its own programs. In other words, it’s the staple of dystopian sci-fi movies which always seem to feature AGI that concludes humans are the biggest threat to the planet and then tries to wipe us all out using its army of killer robots. Funsville.

While this may seem far-fetched, enough highly-respected technology leaders and AI developers are deeply concerned that at the moment we’re not putting in place enough controls – regulatory or otherwise - to prevent this AGI – or even GenAI for that matter - running amok. The more networked and connected our world, and the more decisions we delegate to machines, the greater the danger that errant AI could take over our systems and start doing things none of us predicted or wanted.

Yes, AI promises enormous benefits to Society that are not just commercial – particularly in the field of drug discovery, disease diagnosis and digital healthcare – but it also poses significant threats if we don’t act now to control it. While the necessary regulation is currently non-existent, the good news is that governments and tech leaders are finally getting the message.   

 
Guest UserAI