We all know that AI is now pervasive, and it’s often difficult to tell what’s fake and what’s real. For my part, the only times I use Copilot or Gemini are to acquire images that are copyright-free for publishing here on DCT and my own websites. So today I put them both to the test, just for a bit of fun.
Google Gemini
I have the Google One Standard plan, which I think is the cheapest and gives me 200GB of storage, editing tools in Google Photos, and some access to Gemini. This morning, I asked Gemini to create some images as follows:
Me: Create an image of Lara Croft fighting enemies.
Clearly, Lara is at an advantage because she’s grown a third arm! Very handy when fighting hordes of enemy mercenaries.
Me: Create an image of the widest avenue in Buenos Aires.
It would be lovely to live in a city surrounded by mountains, but sadly, that’s not the case with Buenos Aires, which is as flat as a pancake. Also, if the traffic moved as it does in this image, it would be even more chaotic than it is in reality. Just check out the arrows! Furthermore, the green centre area is in fact the Metrobus route on 9 de Julio.
Me: Create an image of Lara Croft killing a dragon.
Are they fighting together or against each other?
Me: Create an image of a chaotic Buenos Aires.
Buenos Aires has been known to resemble this scene, especially back in 2001. But buses with wheels both above and below would certainly be a novelty!
Microsoft 365 Copilot
Unfortunately, Copilot is very dull by comparison and will not produce any images related to proper names, whether real or fictional.
Me: Create an image of the widest avenue in Buenos Aires.
This is slightly more accurate but also misses out the Metrobus, and also has the traffic moving in odd directions. Also, the quality is nowhere near as good as Gemini’s. I’ve used Copilot frequently, but, unfortunately, it won’t even come up with an image of a fictional character. So then I asked Copilot to create an image of the last British president – the title of my last book, as it happens. However, it refused to do that, saying that the UK is a constitutional monarchy with a Prime Minister. No kidding! So I asked it to use its imagination and come up with a British president standing in front of Buckingham Palace.
Nice try! But since when did Buckingham Palace have a dome? And our man should really be flinging that sceptre over his shoulder or into the crowd, not to mention that his lapel pin looks suspiciously like an American flag.
Next time around, I’ll delve into other AI programs such as Grok and ChatGPT.
—
AI is now so insidious on Google, it is hard to find genuine news items.
Every ‘news’ item now comes with a warning that the pics were made using AI.
That’s not news. That’s art of some sort dressed up as supposed news.
Almost daily I find a new reason to leave the digital age on the table and walk away.
Soon there’ll be enough reasons built up to actually cancel my subscription.
This site is one of a very few I still bother to read.
Agreed, Clissa and I’m glad that you still read DCT!
Thanks Marc
Ai pictures and videos are everwhere now and I notice them especially on “X” (formerly Twitter) where I like to graze to get the latest breaking information. It is getting harder and harder to differentiate fact from fiction when it comes to images, still and moving except in the more obvious examples such as you mentioned (luckily Lara Croft only had three arms because that’s not what most of us used to play the game for !).
I think this will become more and more of an issue unless we have some way to verify what is real and what isn’t. Now we have started using Ai to determine whether something is Ai or kosher.
What do we do when the Ai decides to start deliberately lying to us ?
Cheers
Reg
Hello Reg
I used to graze more on X, Threads and others, but not so much now because of not being able to verify images, as you mentioned.
As for Lara, she has been noticeably compressed in the two new games that were revealed on Thursday…