This is the official Perth blog site for posts, comments, and other contributions about leadership, behavioral finance and economics, and about management generally, as well as other related topics that take our fancy.
Font size: +

LLMs Kaput; Viva la Multilarity!

Words are so boring. We’re going to move to LXMs (non-LLM-based models) which can ditch obsolescent codas to move up into the new good stuff.

Not So Single(sic)

We’ve all been captured by the tempting notion of the Singularity.

Singularity is the final signal of the end of everything, the ultimate resting place for bureaucrats, politicians and wannabe intellectuals. We get there by shamelessly plagiarizing the words in LLMs and improving only one level beyond the trillion monkeys typing since the beginning of the Universe. It represents mental death, or worse.

Now it’s all about AI and the chat GPTs Many out there have hitched their horse to that wagon. How sustainable is it? Does it have legs?

AIs are nowadays based on words. Words in some shape or form have been around a few thousand years, but they are still built on ancient codas. Vendors are copying furiously to steal all the available texts. I think they’re going to run out of runway shortly. They will have exhausted all the available words. In any case not enough intellectual bandwidth for true AGI. What then?

After the Simple-arity

In truth our current AIs and LLMs are only the tip of the iceberg, maybe not even that. In order to achieve true (?) AGI and the fabled Singularity we need way more data, even assuming we can figure out what to do with it. We will need to have a massive collection funnel to get enough. We will also need way more imagination.

We’re going to have to start collecting and interpreting new types of codas and their accompanying codes. These need to have orders of magnitude more types of content and volumes of data. This will need to be unfamiliar so that we can be sure of covering all the bases. Where do we start?

Music probably. Genetics too., a ton of stuff there including proteins, mRNA, other as yet undiscovered biological thingies. Quantum codes likely (i.e. LQMs).

And then there’s some others which will be fascinating including facial codes, emotional codes, what about olfaction, haptic, and so on?

So, moving on to X codes from words. There’s a lot of them. LXMs that is.

Wordless in Seattle

In each domain we will move onto new forms of communication rather than old-fashioned words. In each case we will have to figure out the basic structures and contents of meaning in totally new domains. How do you approach mathematical topology and the codes of dimensional domains? What can you learn in n-dimensions when we have ditched words?

There we will figure out the link between the structures and phenomenological content of the new domains, their types of meaning in new semantic worlds, and how this grants us entry into the assuredly weird mental contents of these domains. Only then will we start to get anywhere nearer to what AGI really is.

The current LLMs might give us some incremental insights, ok fair enough. But what if I want to find out how quantum biomarkers can be used to cure ALS or other neurodegenerative diseases? What about the facial expressions of alien plant-equivalents? How about practical applications of animal esthetics? And so on.

Dread Ends

The AI we are stuck with right now is the end of the road, not the beginning. We should be mapping out a future for AI, non-human or even non-sentient if need be.

We need to find new ways of moving forward with AI things so that we can discover truly shocking, provocative and mind-shattering phenomena that lie way beyond our current imaginations.

It’s not the Singularity we should be aiming for. It’s the Multilarity; multitudinous multidimensional channels of self-conscious, autonomous, thought clouds shooting through not just our planet, but anywhere they can get to, using whatever means happens to be available.

What we need to do is think right past current AI models as being just too simplistic. Towards the Multilarity that lies multiple dimensions beyond. That’s where the LXMs hang out. That’s where we really all need to be.









Stay Informed

When you subscribe to the blog, we will send you an e-mail when there are new updates on the site so you wouldn't miss them.


List of all Perth posts