Nothing in this article constitutes or should be taken as financial advice and should be treated as for general and illustrative purposes only. Always do your own research and seek appropriate independent financial advice.
What is Artificial Intelligence?
Artificial Intelligence (AI) is having a major moment in the sun. From about the middle of last year, various AI apps have been gaining popularity and starting to dominate dinner party conversations, reaching a new high with the release of ChatGPT in November last year.
AI is not a new concept and has been around in its various forms for many years. If you had to call your bank or an airline, you probably had to deal with a chirpy robot on the other side, attempting to have a conversation with you, triage your request, ask for your personal details and, more often than not, drive you utterly mad. The same goes for the pop-up chatbots on e-commerce websites who are always eager to help and sometimes are even half-useful.
Siri, Alexa and the nameless Google Assistant (that a friend of mine lovingly refers to as his “Google Lady”…a.k.a the longest relationship of his life) are also examples of artificial intelligence and have firmly integrated themselves into our lives (even if it is just to update you about the weather).
So, what’s different and why all the rage? Well, there are different types or rather stages of AI. The ones described above are all examples of Artificial Narrow Intelligence, whose focus is on performing a specific (narrow) task or a function, such as search and retrieve data.
However, the latest cohort of artificial intelligence, called generative AI, is based on a more complex type of AI that uses machine learning, meaning it can train itself and learn from large sets of data. The more data it has access to, the more sophisticated it becomes, meaning the quality of outputs improves over time as the AI model gains access to more data to analyse. Generative AI uses these machine learning algorithms to create text, image, video and audio content. McKinsey&Co describes it with a simple example whereby the AI is moving from being able to ‘perceive’ and ‘classify’ an image of a cat to now being able to ‘create’ an image or a text description of a cat.
Examples of Generative AI
This is where it gets fun. Generative AI has gained airtime in our social media and news feeds because to put it simply – it’s fun. Visual AI models and apps like Lensa AI, DALL-E and Midjourney are all examples of generative AI. In case you missed it, those avatar-looking portraits that everyone was posting on their Instagram late last year were generated by the Lensa app. While Midjourney and DALL-E can produce images based on your text prompts.
For example, the image below was produced by DALL-E based on the following text prompt from me: “painting of marie antoinette drinking coca-cola”
ChatGPT, which was released in November last year, is another example of generative AI but this time for text rather than images. Based on your text prompt it can write anything from jokes and essays to computer code, summarise and paraphrase texts and everything in between. So far, some of the more creative uses of ChatGPT that I have heard from my acquaintances included a wedding speech and (drumroll please…) chat-up lines on dating apps, used by a gentleman who struggles to come up with witty opening lines. The latter I found equal parts genius and disturbing.
What does Generative AI mean for the future of work?
All the fun aside, generative AI also opens new possibilities when it comes to work. While some commentators share their concerns about AI making some jobs obsolete, I believe this is a major opportunity. Undoubtedly some roles will change or even disappear as a result of this new technology. However, the same thing can be said about the wider adoption of personal computers which kicked off in the 1970’s and caused jobs like typists and typesetters to become obsolete.
For instance, visual generative AI has the potential to disrupt or democratise (depending on how you look at it) design and content creation. And because a picture speaks a thousand words, here is an example of a candle design I created using Midjourney and the following prompt: “candle in a pale pink and gold jar with a black label, letter A”. It took me the whole of 5 seconds to generate this image.
As with anything, there are two sides to every story. For a graphic designer, this may not be great news, but for a budding entrepreneur looking to start a candle business on a shoestring budget, this creates optionality. Or perhaps, for a graphic designer, this is a way to automate, improve efficiency, lower prices and attract more customers.
Chatbots like ChatGPT will also cause their fair share of disruption by way of greater efficiency. Most likely in the fields of marketing, content creation, copywriting, e-commerce and education just to name a few.
And if you are still not buying it, how is this for a vote of confidence - in January this year, Microsoft announced it will be not only extending its multi-year partnership with OpenAI (the creator of ChatGPT and DALL-E) but also investing US$10 billion into the company.
But while language models can be highly sophisticated, they are not perfect, or at least not yet. For one, they lack common sense, so while the quality of outputs may be high (i.e. how it is written), the content itself may not always make sense and requires validation and scrutiny. Broadly speaking, generative models do not generate original content as such, but rather create new derivative content from already existing data. This means they are not capable of ‘creating’ in the true sense of the word.
Dan Turchin, the host of AI and the Future of Work podcast, in one of his latest episodes, offered two perspectives on how AI will influence the future of work. The first is about the evolution of the workforce into a ‘worknet’, whereby people will be able to move into a portfolio career based on their skill set, rather than working for a single employer for many years. These skill sets will be defined by not only what people are good at doing but also what they are most passionate about doing, resulting in more engaged and aligned workers. The second perspective is that AI-related automation technologies are going to give every employee back at least an hour a week. “That’s an hour that we can be better spouses, friends, caregivers, or pursue a hobby, and that can free us up to feel like we are our best selves at work,” remarks Turchin.
From where I stand, AI models are very useful tools in many professional toolkits. They will create efficiencies, democratise certain sectors and drive technological progress at scale. But they will not, at least in the medium term, replace the work of a human. They will change how humans do work.
Note from the author: GPT4 (the next evolution of ChatGPT) was released shortly before the publication and hence was not included in this article.