Circles
Over the last couple of weeks I’ve counted a minimum of eighteen press releases on the latest LLM developments in my - albeit relatively narrow – AI news feeds.
Added to this the never ending proclamations from every software vendor under the sun about their latest AI updates - creating efficiencies, delivering increased customer loyalty and adding huge value to their (and your business).
You kind of need to prompt your chat agent of choice – to “as a management consultant, please tell me the most important 5 developments in the AI/LLM space this week”.
And that action could been seen as the last step of Dead Internet Theory. I won’t bore anyone or insult anyone’s intelligence with a clumsy explanation of what it is. I mean, with the term Dead Internet Theory you get the idea – even though it’s not a theory anymore.
Here’s a link to some wonderful explanations of what it is. Take your pick.
https://www.bbc.co.uk/sounds/play/p0lbq62j (This one drills down on bots – a connected subject, especially when it is estimated that 50% of the content on the internet this time last year was bot generated – Covid was good for ‘deep state’ bots…)
This theory/reality sits at the centre of how and what our digital selves and (by association) we become – professionally and personally. By way of an explanation for me, just to break down a simple three step process, that I would guess we are all doing on a daily basis.
1. Using AI to analyse, search, check, research, create content, advise, compose. We all use one of…
Chat, Grok, Claude. These are accessing and learning from an internet dated at least 6-12 months old. Hence (with the exception of Grok – too many angles on that decision to bore anyone with now) why they are all signing deals with as many data sources as possible.
Gemini – Google still own search, and index everything. They have that as an – for at least a little while longer – up to date and more reflective source.
Perplexity (my LLM of choice personally – lots of reasons, again not for now ) – have, as with Google, integrated search with AI and recently included a browser (Comet) in their stable. Multiple sources of user behaviour, multiple channels, all up to date.
2. Taking that information and advise and a) Double checking it on another LLM (AI tool) and then b) publishing it in one way or another. Whether sending an email, social post, webpage, message – its all on the ‘web’ and its all indexed somewhere by whatever platform is used.
That platform feeds the information into its own AI model. The content is on the web and indexed by one of the LLM’s in step 1.
3. We repeat step number one tomorrow, aware that our question or task will be supported by AI – which has sourced it’s answer from an AI generated source.
So, professionally we are asking AI for advice/insight from an AI flooded resource. Personally, we are asking AI for recommendations from platforms that have databases flooded with AI generated content.
As I usually find at the end of my ramblings, I have more questions. What does that make us? What does that make our businesses? Do we care?
As usual – John Oliver hits the spot on how some have already given up on their own professional integrity and will lead me into some more ramblings – its 30 mins long but as always, worth it. https://youtu.be/TWpg1RmzAbc?si=EOPTA3EXPs5naFb8