In 1985, my father got his first personal computer at work, when he was about 50 years old. And I remember him saying as if it were today: "I don't have to be bothered about this new invention. My secretary (naturally, it could have been only a woman at that time ;) can learn about it, do it for me and I will keep on working as I always have. Probably it's just a temporary phenomenon anyway." His statement triggered a heated discussion with me where I told him - as a passionate teenager that I was - that he cannot close himself to innovation. Computers are here to stay. And that he should learn how to handle them because otherwise he might lose substantial access to society. It took many months (!) and discussions until he believed me. Today he is happy he did because otherwise he wouldn't be able to communicate with friends anymore or doing his banking - I don't have to tell you how much of our daily activities is solved with digital solutions. Hadn't he dared to believe me in the 80s and finally sit down with the new reality, my dad would neither understand me anymore nor be a full member of this society. He'd be a person without agency, fully dependent on others.
What does this story have to do with ChatGPT? Just like computers and the internet: Generative Language Models like Chat GPT are here to stay. We can criticize their biases and effects on humans as much as we like. They won't go away. To the contrary, they will be used increasingly. At the start maybe more in business. But they will become a major part of our private lives as well.
When we published our study on the question if ChatGPT is able to generate Inclusive Language, we also got negative feedback saying that we shouldn't take this development so lightly; that Generative Language Models do have bias and can be dangerous for this world. And especially, we at Witty Works to whom bias and Inclusive Language is such an important topic, should not be all friends with ChatGPT.
If you want to go more in depth, read our blog post What is inclusive language? And why should I care?
I totally agree: We shouldn't take the recent developments in Generative Language Models lightly at all. To the contrary, we should be super-conscious about the bias it replicates.
And we are also convinced that marketers, communicators, and HR representatives should be educated about this problem and take particular care to use Inclusive Language. For their own success and for their brands, they need to use language that represents today's diversity in society and business, in order to attract and retain customers and talents.
But it's also a fact that in order to be successful, they will use tools like ChatGPT - to write more rapidly, more creatively, and to prevent boredom (because, let's be frank, writing an Easter ad campaign for the millionth time in a copywriter's career is just not inspiring). A recent, non-representative study of FreelancerMap in the German speaking region of Europe found that 61% of the 854 freelancers surveyed use Artificial Intelligence (AI) regularly in their work and 40% of these use Language AI to help with copywriting.
So, I have a choice: Do I accept this fact or do I fight against it because I know that working with Language AI bears a lot of bias risks and potentially can do a lot of damage?
Well, I decided that by refuting the development of AI and diabolizing Generative Language Models such as ChatGPT, I don't solve anything (many thanks go to my co-founder Lukas for the discussions we had). Even worse, by protesting from the sidelines and advocating for Language AI models to go away, I miss the connection to innovation and the real world. Finally, I make Witty Works incredulous in the face of our users and partners. So, instead of disconnecting ourselves from the discourse, protesting the bias and telling marketers not to use ChatGPT, we decided to have their backs.
With our tool 'Witty' we assist users of Generative Language Models to detect the biased wording that may be produced, so that they can take a conscious and educated decision about how they want to use the AI generated content. Witty also assists them to fix the biased wording by giving inclusive suggestions. Based on NLP, Witty is not an AI, but an Augmented Intelligence solution that's designed to improve human decision-making when writing by making aware of bias in the results of Generative Language Models.
With Witty, we want to be a change maker from within, a constructive participant in AIs development and one that can be an influencer and an agent towards less bias. And we can only do this, if we keep "inside of the system" and help monitor Generative Language Models with our own digital solution. With Witty we help authors to combine the best of both worlds: AI generated content monitored by AI on Inclusive Language. That way we keep the agency of holding creators of Generative Language Models accountable and we keep on being a competent business partner of our customers for monitoring the results of AI generated content.
Finally, we have a much bigger, bolder goal. And while some say this may just be utopia, it is our deeply felt mission: We have the conviction that we can change the paradigm of language AI. Along Donella H. Meadows words: "So how do you change paradigms? [...] You keep pointing at the anomalies and failures in the old paradigm. You keep speaking and acting, loudly and with assurance, from the new one. You insert people with the new paradigm in places of public visibility and power. You don’t waste time with reactionaries; rather, you work with active change agents and with the vast middle ground of people who are open-minded.”
We have built Witty to be scalable and to be the helper for the open-minded; those that want to make Inclusive Language the new normal. And if more and more people use Witty to create inclusive content or to adapt the language generated by ChatGPT, Inclusive Language will find its way into the Generative Language Models. The machine learning models will pick it up and changing them from within, overturning the bias they currently replicate. Your support in this endeavor is highly appreciated.
Naturally, the introduction of Generative Language Models is a much more complex discussion than the introduction of computers in the 1980s (by the way, I know that, but my dad at that time didn't). But there are similarities. I am a bit over 50 now, just like my father was then, and though I might also had the thought of just wanting to deny the recent developments in AI and the ethical questions they provoke in me, I know I can't just stay neutral and pull out of the discussion. Why? Because the fact that language AI is here to stay, just like computers, means:
So, just like my father recognized after some discussion-intensive months that he can't deny computers, I won't spend time diabolizing ChatGPT for its bias. Rather, I spend my energy to stay educated, to be part of the development and to make Witty an indispensable whisperer of ChatGPT and other Generative Language Models. Because if Witty is used at scale to monitor and improve AI generated content, Inclusive Language finds its way into Generative Language Models. And that will change the world.
...
If you are looking for a digital writing assistant that can help you to monitor ChatGPTs generated output, try out Witty for free. Witty detects non-inclusive language and suggest inclusive alternatives.