What is the impact of both so far, and as content writers, do we really have anything to worry about?
I’ve never known anything evolve so fast before, at such lightning speed. It’s so rapid that as I started researching the topic, so I could write something down that sounded at least half-intelligent - something else was happening, and more changes were occurring, if I didn’t get it down now, I’d have to start from the beginning again.
Unsplash | Kevin Ku
I suppose I shouldn’t be surprised, since OpenAI introduced ChatGPT back in November last year its evolution has been rapid. And now we have Google Bard. There’s been many a heartfelt predictions on where it’s going and what its impact will be on our lives. People have predicted it taking over many jobs and the end of original writing.
And writing is where the hurt is. AI can already write books, (it’s written at least one children’s book), code, write articles, answer questions on strategy and even do maths.
There’s a wave of panic cascading through the online writing community, which just happens to coincide with one of the worst economic cost of living crises we’ve ever known. And it’s predicted to last until at least the second half of this year (2023), possibly into 2024. with 75% of the bottom 20% of low-income households in the UK alone going without essentials.
There have been massive job losses across the US and UK, and this is especially acute in the tech sector where different theories abound on what the cause is. Most point towards an over-enthusiasm for hiring during the pandemic - an expansion that turned out to be unnecessary. But the economy clearly doesn’t help, business is slow, slow to hire and even slower to part with their money.
We shouldn’t underestimate AI’s capabilities, but we shouldn’t overestimate them either. AI is created by humans and it is what humans do or not do that we should fear most. AI does not have independent intelligence of its own - yet. What it does have is an unconscious bias which they get from humans - we should definitely fear that. And at the moment, a propensity for producing inaccurate information and emergent abilities -
“Emergent abilities are skills that suddenly and unpredictably show up (emerge) in AI systems. In the past year, emergent abilities have gained significant attention as intelligent machines acquire more skills and our understanding of what’s happening inside them grows more opaque.”
Those final few words are the most worrying. However….
In a recent piece of research from Cornell, emergent abilities may be little more than a ‘mirage.’ Examples of some larger language models (LLMs) are those we know the most about, such as GPT, LaMDA and PaLM. The emergence of extra abilities has, as you can imagine, garnered a lot of interest. ChatGPT 3 may have been the first to spot emergent abilities and it has become increasingly obvious that experts are finding it hard to predict which abilities are going to spring up next.
These sharp and unpredictable abilities have been described by Cornell as little more than a mirage, and argued that it is nothing more than “sharp and unpredictable changes in model outputs as a function of model scale on specific tasks,"
*A copy of that report is here - https://arxiv.org/abs/2206.07682
In the meantime, change continues with GPT4 in the form of more up-to-date information, up until 2023 rather than 2021. This means it can easily compete with Google Bard, Chat’s rival, which can write on your behalf or answer questions with relevant information right up to and including this year.
Unsplash | Google Deepmind (yes I know!)
However, like GPT3 it can still have hallucinations and produce inaccurate information as fact. These aren’t hallucinations that we might recognise as human in nature, the word is not used as it would be in psychology, instead, it is used to describe what can happen when it produces false information due to an internal error (if you can call it that).
But, so impressed are certain organisations with GPT4, Forbes reported earlier in March that-
“On Twitter, [they’d seen] many developers testing out GPT-4 with requests for it to code old-school arcade games. While GPT-3 often provided buggy code that needed re-prompting, GPT-4 in many cases gives perfectly working code to start.”
Tyler Weitzman, cofounder and President and Head of AI goes on to suggest that while -
“ some see it as potentially concerning for programmers, my company is considering allowing candidates to use it in interviews since they'll be expected to utilize it on the job.”
Early Bard didn’t seem to produce the same enthusiasm according to Business Insider back in April when Google HQ was required to test the chatbot on behalf of their employers, with one employee calling it “cringeworthy” while another went as far as to call it a “pathological liar.”
Some argued that Bard could be dangerous after giving scuba diving advice that could have resulted in death. Apparently, comments like these were taken from screenshots Bloomberg got hold of from internal discussions.
There are still errors and obstacles to overcome for these most well-known chatbots, but it doesn’t stop people from using them. This has prompted some businesses to stipulate that any content writers they hire must adhere to strict rules on writing - that do not include the use of ChatGPT or Bard. Some are even going as far as to prohibit use completely by using AI detectors to ‘out’ anyone who may have tried to use either one covertly.
Unsplash | Andrew Neel
In an unexpected move from Apple, employees are now prohibited from using chatbots. Their reasoning is that by using these too frequently, confidentiality information could be leaked or even purposely collected by rival firms. And it’s not just ChatGPT it’s been warned to steer clear of, it also includes GitHub AI programme assistant, Copilot.
Overreaction? Possibly, but all interactions between users are stored on OpenAI, and what’s more, these conversations are used to train their systems which can then be inspected by moderators.
As writers, should we be worried?
The way things are right now, yes. It’s quiet, writers seem more worried now than they were during the pandemic. Copywriters are feeling the silence, even the amazing and brilliant ones, here in Europe especially it’s bad. But there’ll always be someone who’s doing okay because there’s always someone who can make money during a recession.
The general argument is that as long as we prove our uniqueness against the fake of AI, we’ll be okay because eventually the hirers and firers of content writing will soon see through the veil of AI greatness and realise how essential humans are. But the truth is that there have always been content/blog providers that prefer to do it on the cheap long before AI came along, and AI will simply make things simpler and easier - and save money.
Cut out the middleman that is Upwork and Fiver and just let AI do its magic. That could go one way or the other, but it’ll take time before any of us see the impact, and it’s difficult to predict how it will affect the writing industry because it’s evolving so quickly.
But there’s hope. Content writers have argued the importance of uniqueness, the uniqueness of being human as opposed to what AI can do. But the stark reality is we won’t be unique for that much longer, because if AI continues to make the strides it’s making now, it will be able to do pretty much everything we can anyway before long.
What we should be striving for according to Samuel Arbesman in his article “AI, the God of the Gaps, and Our Quintessential Humanity” is to be quintessentially human rather than unique, because it is what makes us unbeatable in comparison to AI. AI may be able to do what we can, but it can never do it in the way that we do it. We are original, spontaneous and eccentric. As Arbesman argues -
“Can AI do boring and crappy jobs? Hooray. I didn’t view that as quintessentially human anyway. Can AI write poetry? Wonderful, so can a non-trivial fraction of other human beings. Each of us must determine what is our quintessence, whether it’s spending time with family, writing blog posts, gardening, or playing chess. Just because these are not necessarily uniquely human behaviors doesn’t make them any less quintessentially human.”
Never before has content writing been so threatened, but at the same time never before has it been on the brink of something truly amazing. Now, it’s time for those of us who write in a way that no one wanted before because we couldn’t squeeze ourselves into the blank box of mediocrity. A time to shine for those who have a particular style of writing that’s instantly recognisable, and a big hurrah for those that value voice of customer research - something AI cannot replicate when it comes to copy that converts.
This could be your time, our time - you can either stay and fight or give in to the machine. It’s your choice, and there will be a rocky ride ahead, but before you decide to throw in the towel remember that no one can be you, and AI can’t be, in Arbesman’s words “quintessentially human.”
We have it all to play for.
Kommentare