I like this post about ‘impressionist blogging’, mainly because it gives voice to a sense of how using generative AI as a tool shapes the work it serves. I’ve tried, on and off, for the past year to see how generative AI could be useful in my workflow. I quickly ruled out any copywriting that would see the light of day. The voice veers into a lyrical uncanny valley. Research and learning about new things are out too. If I still have to fact-check everything, why not just do the reading in the first place and get the benefit of serendipity? I settled on one use case: marketing planning. Confronted with a time-crunch, an LLM turned out to be a shortcut to hard thinking about how to position a product and who the potential customers would be.

“There are certain tasks where technology can be a lever to help us be more productive. And there are other tasks that beg to receive the attention of human thought and emotional energy.”

https://blog.jim-nielsen.com/2024/impressionist-blogging/

The output of that exercise was genuinely useful, but it is not something I want to make a daily part of my working process. I’ve worked in PR and marketing for the past 14 years, and I like to think have the competence to produce a communication strategy or marketing plan based on my own research and understanding. I have been involved in plenty of marketing and comms campaigns, and again, I like to think the accumulated experience from that is part of what makes me valuable as a professional. Those experiences and what I took from them are unique to me and can’t be synthesised by a large language model.

This year’s Edinburgh Book Festival had an excellent panel session tackling AI with Shannon Vallor, Baillie Gifford Chair of the Ethics of Data and AI at the University of Edinburgh and author of ‘The AI Mirror‘, and Anton Hur, author of ‘Toward Eternity‘ and translator. Anton touched on the impact AI can have on translation, with a growing expectation from potential clients that machine translation can replace the work that a human translator does. I think it is safe to say that he didn’t agree with his replacement. Later in the day, in a session with Emily Wilson, Professor of Classical Studies at the University of Pennsylvania, who recently translated Homer’s Iliad, that was reinforced for me in the discussion of how she had been faced with choices of poetic metre and word-choice in the translation. AI can approximate a translation, or perhaps a sort of transliteration, of a text, but the output will always lack the soul of a human translation. A machine cannot experience or understand emotion or irrationality; a human can.

I still don’t know how generative AI will work its way into my life. I’m far from technophobic. For all the risks that it poses, there are myriad interesting possibilities. But perhaps they are not to be found on the path we are currently walking.