26 Jun Gen AI Rage: Why people are getting poor quality responses and getting mad
Introduction
The rise of generative AI tools, and the hype surrounding them, is leading to skyrocketing use, but also growing misuse and unrealistic expectations. Poor outcomes are leading to what I have labelled “AI Rage”, like road rage but fortunately no collisions or fist fights.
AI rage as a response to ‘AI ineptness’ is a parallel to Road Rage because it is triggered by another entity (the AI) not behaving as you would wish.
However, whilst it’s hard to influence the raging idiot who just cut in, you can quite easily, and dramatically, improve the outputs from most Gen AI tools.
Read on to find out how… or remain annoyed and throw the baby out with the bathwater.
Your call.
The Importance of Proper Instructions
Prompts are instructions given to generative AI to produce desired content. Poor prompts lead to poor results, period.
For instance, asking a generative AI to “write a story” without context will yield vague content. Instead, specifying “write a 500-word story about a child’s adventure in a forest, including 3 fantasy characters, during a rainstorm, leading to a life epiphany” results in more useful output.
Equally a longer prompt without clarity is no better, in fact it’s probably more confusing and more difficult to respond to. For example, uploading an entire article and as for a cover image, is unlikely to work well, but clearly describing the kind of image you think represents the piece will be much more fruitful.
In the case of prompting the old adage applies: Garbage in Garbage out.
Common Misconceptions About Generative AI
Another misconception is that AI can generate perfect content without human oversight. As discussed in the prompting example, this is not the case. AIs have no context and whilst they can be creative they need a ‘tight brief’.
Some of the large providers are attempting to build features that create context for AIs (e.g. OpenAI Assistants and ‘Memories’), but right now if you are using a vanilla chat interface you will need to provide all the context.
Next misunderstanding… expecting AI to replace all aspects of human creativity, it’s just not realistic. Effective AI use requires collaboration between human input and machine processing. You can think of AI as a bigger engine, an accelerator or an incredibly fast but nieve assistant. If you are expecting the AI to lead or be in charge, you will be disappointed.
Keep an eye out for Agentic AI though – when successful this will be a step change in the capability levels and applicability of Gen AI.
Using the Right Tool for the Job
Generative AI is often embedded into tools where it’s not appropriate, or it is not well integrated. Everyone is jumping on the bandwagon and this is effectively supercharging crappy tools.
For example, using a drawing application that can only wireframe and expecting to see The Garden of Earthly Delights as an output is misguided.
Once more this is poor use of Gen AI, but in this case it is like putting Tiger on a crazy golf course with an umbrella for a club.
Conclusion:
Generative AI holds tremendous potential when used correctly. By refining your approach and understanding the tool’s capabilities, you can achieve better results and enhance your productivity.
By getting annoyed and saying so on social media, people are stoking the fire and missing an opportunity.