Which is the real image? The man with the cars or the girl? (Remember me asking a similar question back in 2022?)
Okay, too easy. The girl has an extra finger .
But which image hit you the hardest when you first looked at the pictures?
When I saw images online from the Valencia flash floods, my first thought was 'Fake'. Other people were suggesting the same thing on social media - already there's widespread mistrust of images. However, when I did some cross-cutting searches, it looked like the Valencia images were for real.
I was feeling particularly sceptical about truth on the web this week as my Google searches now provide an AI summary at the top of the page. Google launched these summaries in May in the US and rolled them out in NZ on October 29th. You can't get rid of the summary on the search page, although if you take an additional step and choose the toolbar option 'Web', you don't see the summary.
These AI summaries annoy me intensely. We know Large Language Models do not look up and interpret information, they predict the next most likely word. As a result, although LLM material may be accurate, it is sometimes highly inaccurate and there is no telling the difference – the LLM's have no internal check and balance for truth in their outputs.
When I want to verify search information, and that's most times I do a web search, I use a sweep of search hits. I know this is also a flawed method – search engines have their own, secret, ways of prioritising the order in which information appears and which websites appear at all. However, some comparison is better than nothing. I believe Google's AI overview will rapidly be the information of choice for most people searching, rather than any actual website. I'm already finding the Google summary hard to avoid. Try as I might, I flick my eyes across the overview while attempting not to. We will all quickly be lulled into thinking we have got a summary of information from the overview, when we have not.
The other driver of my scepticism about reality on the web this week was reading an Economist article about the role of influencers. I thought influencers are for the Millenials but, no, it turns out 60% of GenX (my generation) in the US are sometimes guided by social media influencers when making a purchase and even 35% of Boomers are influenced. So influencers have...influence. The Economist article focused on how influencing is becoming a niche game – it used to be there were a few big-time celebrity influencers who everybody wanted to be and everyone wanted endorsing their products. Now companies are looking for more targeted influencers, who have dedicated followings from a niche market. This means there are more influencers each making less money; only 4% of the approximately 50 million influencers (or 'creators' as they prefer to be called...creators of what?) globally make more than $100,000 a year.
And, watch out human influencers, AI influencers are on the rise and are already displacing you. In researching influencers (I refuse to call them creators), I read about Aitana Lopez, a 25-year-old, pink-haired fitness influencer who is an AI Creation of 'The Clueless'. The Clueless are "...not just an AI modeling agency; we are visionaries on a mission to redefine the world of influencers. Our goal is to transcend the conventional realm of modeling and lead the way in influencing, ushering in a new era characterized by authenticity, impact, and deep meaning. We are dedicated to meticulously crafting AI models that go beyond mere appearances, capturing the essence of diverse personalities and experiences. " So far The Clueless have three AI models, all young women...not so diverse.
Aitana has over 300,000 Instagram followers. It turns out she's a 'baby' influencer – Lil Miquela has 2.5 million Instagram followers. She's a 19-year-old Brazilian-American model and singer who, of course, doesn't actually exist. Lu Do Magalu has a stellar 7.2 million Instagram followers – she's also Brazilian and endorses Burger King, Red Bull, Adidas, Samsung and McDonald's, amongst other products.
Here's a top ten list of AI influencers – all of them are women. Here's another top 10 list, which has two men in it. Interestingly, there's a lot more diversity in skin colour and apparent ethnicity in these lists than there is in gender or age. Maybe Boomers don't care if influencers look like them?
As much as I'm not going to particularly mourn the loss of human influencers, I'm feeling there's an insidious shift rocketing through our world that we're not even fully aware of. The influence of AI is mounting (let's not get into a debate here about how intelligent these AIs are, or what AI really means). The influence of the non-humans who are taking over human roles (and not the labouring roles, as I wrote about in AI Frontline) is propagating as rapidly as avian bird flu is rocketing through bird and mammal populations across the planet. Okay, I won't go there – or the US election – there's only so depressed anyone wants to be on a Saturday night.
We are supposed to be comforted by the hypothesis that AI influencers can't replace human influencers – on the basis humans provide emotional connections and relatable experiences AI cannot replicate. I find this hard to believe. We experience emotional connections when watching fictional movies or reading books and all those 'people' aren't real. However, up till this point we have known they aren't real.
So, does it matter if you can't tell who is real and who is AI in the world in which you operate? I don't know. Do you?
27 Aug 2022 Excessive Complexity Will the future be dominated by AI because we have created such a complex world that humans can't cope with it?
25 Nov 2022 Democratisation or Theft Is AI democratising the process of creativity? Is it stealing creative works? Will AI creativity mean humans become less creative?
16 Dec 2022 AI Advances My first blog referring to ChatGPT. How good are large language models, considering they are still babies?
31 Dec 2022 Recognising Faces Can you tell a real image from a fake? AI can't, but it can make fakes. And it can recognise your face. Scared yet? Should you be?
28 Jan 2023 AI with Purpose How I used DALL-E, an AI image generator to create a logo for my publishing company to be able to publish my first novel.
11 Mar 2023 Is AI Intelligent? My blurb was, "Are Chatbots like ChatGPT insulting people? No. Because they wouldn't know an insult if they wrote one." Now I'm no longer sure that's true – LLMs can already explain why a joke is funny.
20 May 2023 AI at Pace Pi is going to make us better humans. Really?
13 Jan 2024 Am I an Anachronism? Will AI make creatives redundant in the broadest sense of that word?
16 Mar 2024 AI Frontline Why is AI taking over the creative jobs where we imagined robots doing our drudgery?
27 Apr 2024 AI Everywhere How fast is AI developing? Faster than I, or Geoffrey Hinton - the 'father of AI' would have believed. Here's a story of AI chatbots replacing friends that shifted from fiction to fact in less than 5 years.
blogger
traveller