Playing with machine learning is both fun and inspiring, also for people from Hallingdal. Let the creativity flow – generate a picture of a swarm of monkeys flying over a mountain lodge at night. Fun! But there is a «but» here...
Playing with machine learning is both fun and inspiring, also for people from Hallingdal. Let the creativity flow – generate a picture of a swarm of glowing bananas flying over a mountain lodge at night. Fun!
«Give me bullet points for a talk on sustainability work in Flå municipality» – voilà!
«Write a blog post about performance culture in speed skating, and a sales text about our charming mountain hotel» – off it goes!
This article was first published in Aftenposten and at Studentersamfundet.
What have we learned after a good year with tools like ChatGPT at work? Can we trust the machine-generated text? (Here, Google’s advanced language model suggests I should use the «word» machine-fired.)
Did I not read something about the language models being inaccurate, and in some cases lying?
Why does DALL-E generate images where people have six fingers, or are missing a foot or an eye?
Perhaps it is time to reintroduce critical thinking – a highly human ability.
Do we really believe that a model generating text or images based on training and probability is capable of delivering communication that truly hits home and engages people?
Communication is first and foremost about empathy and a deep understanding of who you are talking to. What do you know about the audience? Attitudes, opinions, habits, financial resources, family and work situation, interests, group belonging and desires – how do they live? Is there a way to phrase our message that is both relevant and sparks interest?
Communication is first and foremost about empathy and a deep understanding of who you are talking to.
Unfavourable intelligence is blindly trusting that machine-generated (machine-gunned) text and images can replace the human touch. Unfavourable intelligence is a lack of commissioning competence – when the person assessing the machine-generated content lacks the skills to quality-assure it. Because there are worse things than people with six fingers.
Everything we publish, whether it hits the target audience or not, shapes how we are perceived as senders: brand, reputation, and our business’s opportunities now and ahead.
It is possible that generative AI will one day be able to deliver based on an interpretation of every conceivable human emotion broken down to a geographic, psychological and sociological micro level – but we are not there. Not even close.
I am rooting for critical thinking. You could start with the term artificial intelligence itself: could it be a marketing term designed to create hype and attract investors? It certainly sells better than dull words like machine learning and statistical language models.
Yes, I am myself a fascinated power user of these tools. The point is to try to understand what they can and cannot deliver. The point is that all communication starts and ends with humans. For now, I remain opposed to the uncritical use of machine-gunned content.
