For the past 18 months, just about every social feed and newsletter has had a common obsession – GenAI. It started as fascination, the wow factor. Can GenAI really do that? And over time the hype has only gotten bigger: Steal these prompts! Scale with these hacks – all with the insinuation that without AI you’ll never keep up!
The question is: what are we trying to keep up with? Industry trends? Competitor tactics? Tech advancements? Newsflash, these challenges are not new. And the capabilities and impact of Gen AI that everyone is so enamoured with? Turns out, instead of providing the competitive edge everyone is angling for, they’re often just adding to the noise.
There are specific applications where AI is offering some advancements, but in terms of GenAI such as Chat GPT, for the most part it’s just creating more generic content in more of the same generic style, and more regurgitation of what already exists.
While AI advocates seem to be standing firm in their beliefs that the future is all about AI, there are others shaking their heads. “I’m not so sure about that.” Evidence of programming biases, factual inaccuracies, and other inconsistencies are coming to light. The idea that AI can sort information faster and come up with answers is being challenged – because more often than not, AI is not always getting it right.
For images and written content, the lines have become very blurred. Some people claim that they can spot an AI generated post miles away, but even the common tells that people claim to know aren’t a guarantee that it wasn’t written or created by a human.
What makes content great? That’s always been subjective – now AI is training on specific types of writing or image generation. In that training there’s little understanding of nuance or even context. It’s simply something programmers don’t even think of. Until, that is, the outputs become a comedy of errors.
Sure, it can be amusing, but it’s also a problem because as AI capabilities advance, they’re doing so with those biases and inaccuracies embedded into their programming. This is one of the reasons that there are calls for AI development to slow down or pause. It’s not that people are against advancing AI, but that they see the flaws being augmented if they’re not addressed. If inaccuracies continue to grow, the benefits will diminish and so too will the investment value in AI technologies.
For all the excitement of what AI can do, at the end of the day it can only do what it’s told to do. Even with agentic AI, there are parameters that define what decisions it can make and how. More and more people are recognising that while there may be a use for AI, it’s glitterball attraction is just that – a reflection of inputs, scattering them as outputs to the world. And not always great outputs at that. Has AI lost some of it’s shine? Perhaps it needs to polish up it’s act if it’s to win trust back.
