"ChatGPT is a breakthrough in natural language processing, allowing machines to understand and generate human-like language with remarkable accuracy. It represents a major milestone in the development of artificial intelligence and has opened up new frontiers in a wide range of fields, from customer service to creative writing.” - ChatGPT.

Image generated by DiffusionBee with the prompt “inside a classroom, lecture hall, university, graffiti art on a blackboard, dystopian, neon glowing lights, sharp focus, photorealistic.”

“ChatGPT-sjokket”
To get a second opinion on this chatbot I attended the 30th March 2023 Fagdag with X-IMK, titled “ChatGPT-sjokket: Muligheter, utfordringer, risiko?” This was an open lecture style event on the opportunities, challenges, and risks of so called generative AI’s such as ChatGPT. UIO professors, Petter Bae Brandtzæg and Taina Bucher, spoke at the event alongside fellow academic Morten Irgens. Developers at the Norwegian publication VG, Johannes Gorset and Edvard Høiby, gave insights into how AI tools have already begun to change the way in which we work.

What is it?
Up until the publicly lauded release of ChatGPT in November 2022, AI tools have generally been used in very specific ways to accomplish individual tasks. Analysing medical images for cancers, facial recognition at border entry points, personalised recommendation systems in social media - AI has been doing these things for years, with each problem requiring its’ own tailor made algorithm. And that’s what makes ChatGPT different, it’s more versatile than anything we’ve seen before.

It’s a generalist, that among other things can code, write content, translate between languages, and tutor students. By learning from a dataset that contains billions of webpages, ChatGPT has learned to mimic human language, and arguably human creativity. But it’s here that we come across some limitations of these so called large language models (LLM).

What isn’t it?
Fundamentally, LLM’s have no common sense reasoning skills. They don’t know right from wrong, which is why they can do genuinely insane things like falling in love with the user and plan for global destruction. The industry term “hallucinations” is perhaps a charitable way of referring to the hate speech, offensive content, and violent threats that ChatGPT has been seen doing. The only morality the algorithm knows is the oftentimes inadequate safeguards imposed upon it by its’ creators, and is only as ethical as the humans who program it. It’s simply not a true artificial general intelligence (AGI).

The incredibly natural and convincing way of writing that has so quickly propelled ChatGPT to such incredible heights (100 million users in 2 months!) has also driven a phenomenon repeatedly mentioned by the lecturing experts. Fear hype. We’ve blasted past the Turing test with such immense speed that it’s easy to think we’re approaching the ultimate goal of AGI. But we’re not. For all the doomsaying from prominent figures such as Elon Musk and Steve Wozniak, AI is in fact nowhere near to being an existential threat to humanity. That doesn’t mean we shouldn’t be worried though. The rapid pace of change in both development of AI technology and the speed of it’s integration into society is something we should be paying close attention to.

Should we be scared?

“I guess the thing that I’m most worried about is, if we compare [ChatGPT] to the industrial revolution…we had 40, 50, 60 years to adjust and understand it, but the thing with technology is that it improves faster and faster. This time it’s not gonna be 40 years, it’s gonna be 4 years, and are we gonna be able to adapt?” - Johannes Gorset - Director of Engineering, Schibsted

While the broad applications of ChatGPT are very impressive, it’s important to remember that even seemingly magical AI tools are only that - tools. They’re undoubtedly going to transform the way we work, but that’s not a new concept to us. For the longest time we worried about how robots would take all the blue collar factory jobs, the only difference now is that it’s the white collar office worker who's job is about to become automated. Technology has eased the burden of physical labour, and now it’s doing the same with cognitive labour.

That the most tedious and boring parts of someones job can be done faster by a machine isn’t a bad thing. I used the VG-developed Jojo Transcribe app to save myself some time when going through interview recordings, and DiffusionBee to create the images for this article. These tools are already proving useful, and they’re improving at an increasing rate. I’m genuinely excited to see where we’ll be this time next year.

We’ll be OK, probably
The general feeling throughout the lectures was one of anticipation. AI development still has a long way to go before it reaches maturity and the technology will undoubtedly continue to become integrated ever more closely into peoples lives. By 2025 it is predicted that over 90% of content on the internet may be AI generated. That’s going to be a massive paradigm shift for us to navigate, especially when it comes to controlling the spread of misinformation (keep the 2024 US presidential election in your periphery).

We can’t know what the future will hold, but generative AI will play a massive part in it. Pandora’s box has been blown wide open, and for all the attempts to regulate it, things will never be the same again. I for one, am excited about the opportunities this tech will provide us. In such a rapidly evolving world, anything less than an enthusiastic willingness to adapt to change will put you at a disadvantage. Embrace the change, or risk being left behind.

How Generative AI is Poised to Change Everything