AI, ML, ChatGPT; Here comes the stuff.
If you haven’t heard of AI / ML / ChatGPT / CoPilot or any other flavour de jour, then it’s likely you’ve just returned from some two-year journey along the fault lines of 3G / 4G / WiFi coverage.
Everyone is very excited about the possibilities this new thing is going to help to draw out. Personally, I’m relieved that I can use ChatGPT to write out my Excel formulas.
Excel is the fungus of financial and data planning; you’re never sure when it took root, but before long, you’re spotting clumps of .xls files all over the place. Despite how much time I spent in, on or around Excel, I’ve never invested the time to drive it properly (I can remove this from my to-do list now).
To summarise, ML helps you deploy your time more succinctly to tasks which suit them, one of which is a rendition of visual or textual creation.
This is a two-way street because it takes less time, which means output is increased.
The internet is big, super big, so big in fact that the contents of it are used to create the fuel which fires up these ML engines. The giant blob of data represents most of what we know and a whole bunch of stuff people make up.
Aside from the bunch of stuff which people make up for political or personal gain, there’s also a lot of anger floating about. People are motivated to complain when something goes wrong, isn’t the right flavour, or strikes a bad chord. The place they go is the internet. As a collective, I would probably guess the sentiment of the internet is a bit grumpy.
So will ChatGPT or its children start life a little off colour? Perhaps, or maybe, we’ll juice them up to avoid making too many negative connections around the fallibility of humanity.
Let’s get back to the point I’m trying to make. All these new capabilities are going to empower us to create more stuff. Lots and lots of stuff.
The problem with generated stuff is that it will lack personality and context. Because ML(ChaptGPT) is based on the average of everything (Actually a probabilistic matrix with context, i.e. a transformer model), what you create will be the middle ground. It’s not going to be poor; it’s just not going to be great.
Great is making new connections or insight; great is your personality beamed out over time so that people start to get to know you and want to know more.
The overall impact on the internet will be an influx of more average content added to the existing heap. With such a proliferation of average material, discovering noteworthy content will require a novel approach. An AI assistant could potentially be useful here in filtering out generated content and directing users to fresh and intriguing internet creations exclusively crafted by humans.
In summary, then, choice and variety will be reduced because finding wood for the trees will become much harder. Those who drive the tools the best will drown out those with the more original thinking.
I sound largely negative on this whole kit and kaboodle; I’m not negative about ML and AI; they’re going to help us get better as a human race. We’ll find new cures to complex problems, and we’ll remove some of the grind from jobs and daily life. I am more negative about its short-term impact on the internet.
Thank you for joining me, friends, on this outing, a glancing blow to ML-generated content; it’s been nice to see you again. Until next time, over and out.