The Future of Generative AI and ChatGPT
AI is like marmite; you either love it or hate it. There’s no disputing that people either have lots to say on AI or are just begging that no-one asks the question. Sharing their take on Generative AI and ChatGPT is Media Manager, Natalie.
Until recently, ChatGPT isn’t something I can say I’d properly looked at. It felt a bit like information overload with the amount of people talking about it or the constant innovations being reported. Now I’ve done some investigating and listened to different takes on AI as a whole, as well as ChatGPT in particular, I’ve come across some really interesting points of view.
Not every business has an unlimited pile of cash to do extensive research, field testing and customer segmentation. This is where ChatGPT (and other ever growing tools) can help – and I say ‘help’ specifically as it’s not going to do all the work for you, and, if you let it, chances are some of that work will be wrong, outdated or just misinformed.
A recent study (‘Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence’, before you ask) gave two groups realistic 20-30-minute professional writing tasks. One group were able to use ChatGPT and the other group were not. The study found that those that were able to use the tool were more efficient, as they had less heavy lifting to do and could spend more time on strategy and editing. Using ChatGPT with the right prompts gave them the first draft or starting point on which they could then build from, leaving more time for the magic and human inspiration.
The outputs from ChatGPT can be very dry, biassed and even arrogant in tone, which is where the need for editing and reviewing comes in – humanising and refining the output. It’s by no means a perfect tool and has many challenges, including:
- It’s not good at quantifying its findings
- It has an inherent cultural bias and is trained largely on a Western view (it also answers differently based on language used!)
- It’s only trained up to mid 2021, so current trends don’t factor into the outputs it produces
- And the big one for me – it lies. (Check out Varghese vs. China Southern Airlines Co Ltd). This just highlights that you still need to understand the field in order to validate the output. If you knew nothing about what you are asking, then you wouldn’t know if the answer was correct!
AI and machine learning are nothing new but their growing ability, functionality and availability are what’s drawing attention right now. The phrase I hear over and over is “AI won’t replace you, but someone who knows how to use it might”.
With growing computing capabilities, AI is driving us towards “hyper-automation”, where manual work will diminish but humans will continue to be the ones that design and supervise systems. Right now, people are talking about the possibility of AI taking jobs – but really it’s just enabling us to automate dull/mundane tasks that no one really wants to do or that take too much time when that time could be spent on more productive tasks.
To summarise, I think the most interesting and true sentiment on AI is the same with implementing anything new; people need to be onboard emotionally. You also need to work to remove/reduce their fears so that they are able to see plans positively.
Technology is only an enabler. People will always run the technology, and therefore remain the most important part.