Chautauqua 3

 Of dogs and thunderstorms....

The Liminal Space

I have, for some time, been quietly obsessed with the concept of the Liminal Space. Most of my photos over some years have veered towards expression of the aesthetic. A liminal space being defined as follows (from above link):

True to its etymology ("liminal" being derived from the Latin word limen, meaning "threshold"), the concept of a liminal space encompasses physical spaces that, due to their function, are transitional in nature - hallways, waiting rooms, parking lots and rest stops are the archetypical examples of such places. Liminal space aesthetics relate to the unique feelings of eeriness, nostalgia, and apprehension people report when presented with such places outside of their designed context; most notably, their function as intermediary points between origin and destination. For instance, an empty stairwell or hospital corridor at night might look sinister or uncanny because these places are usually brimming with life and movement. Therefore, the absence of external stimuli (such as conversations, people moving around, or any kind of social dynamics) creates an otherworldly and forlorn atmosphere.

Chautauqua 2.

I had the following chat with GPT4, using the Chat With PDF plugin. GPT4 was able to use that plugin and carry out analysis of the implications of the paper concerned. 

Chautauqua 1.

Musing on the prescience of E.M. Forster's The Machine Stops, the conditions as we enter the Singularity, and riding the wave. 

For what it's worth, I have changed my opinion on the utility of the concept of Moloch, but this discussion took place over a month ago.

Is GPT4 Useful At Work?

Is a given system useful for a given task? This calls into being the following chain of reasoning. 

Purpose -> Criteria of Acceptability -> Is it useful?

So are Large Language Models as capable at work as a human? Are they useful? In addressing this I'm not going to make public what I've done for my employer, suffice to say it is potentially useful for us in a wide variety of roles. Instead I'm going to approach it from a tangent.

Is GPT4 just a Parrot?

"GPT4 is just an Autocomplete on Steroids!" So goes the cry of pseuds across the internet. 

Let's see.

Chautauqua 0.

I've been very busy, hence disappearing for months, I am now preparing a series of posts about the ongoing singularity. However I've been using GPT3.5 and 4 as an assistant at work and am eagerly awaiting the incorporation of GPT4 into Office 365 as the Office Co-Pilot.

Anyway, here's something I just tried with both GPT3.5 and 4, I am currently using both at work, but doing so with awareness of their capabilities. Typical tasks include pricing up quotes and checking purchase orders against quotes, GPT4 excels at such tasks.

The Machines will defeat us, even if war is not their aim.

In my last post I made a throw away comment in relation to the subject of climate change: "...nothing exposes right-wing ideological capture more than the subject of climate change, it's their version of the gender ideology clown-world occupied by the left." This ideological capture is most commonly known via the non-playable character (NPC) meme.

Does Flamingo Understand?

Ten years ago Andrej Karpathy lamented the state of AI vision research. He is now a lead on Tesla's autopilot vision team, probably placed better than anyone to have revised his opinion. Anyway, this led to a Deep Mind researcher, Roman Ring, to use the photo cited by Karpathy in a little experiment.

No Reason to see Personhood in LaMDA

I've been doing an awful lot of thinking about Lemoine, LaMDA, and Google. I remain open to there being something like sentience in LaMDA and GPT3, because there seems to be evidence of intelligence and understanding in both systems configured as chat-bots. However I do not think the evidence goes as far as a sentience potentially requiring consideration of personhood.

Down the Sentient AI Rabbit Hole (Again)

As reported in New Scientist and elsewhere, a Google engineer has been suspended for going public with his opinion that Google's LaMDA transformer model is sentient. The experts say he is mistaken, are they just falling foul of Clarke's First Law?

Do Diffusion Models Dream of Electric Sheep?

DoodleChaos on Youtube has used a diffusion model from Google, Disco Diffusion v5.2, to produce a music video. This avenue of AI is all new to me, never heard of this system, but watching it is an astonishing experience.

Agents In Society

If you had a super-intelligent General Artificial Intelligence and you asked it to produce as many paperclips as possible, the results could be catastrophic. It could destroy the Earth in search of the goal of optimising the manufacture of paperclips. This is an often used analogy designed to illustrate the dangers of poorly defined goals in artificial intelligence (AI) and machine learning (ML). However this paradigm of analysis has wider utility in consideration of society.

What is the entity that is 'I'?

I've been pondering on a matter since the 1990s when, amongst other things, I was evolving neural networks on a spare computer. And now, decades on, and much reading since, I think I can explain...

Making Gods is a Dangerous Endeavour

In this post I just want to go over some of what the rather limited current Artificial Intelligence agents are capable of. This is going to be rather cursory and fast, I don't have the time or inclination to write a book. I'm sure you don't have the time to read it anyway. The comments section is there is you want to ask me to back anything up.

Anthropomorphising

The previous two posts contained annotated transcripts of conversations between myself and GPT3 Curie, a 9 billion parameter version of GPT3. When I first came across GPT3 operating when configured as a chat bot it was in the video presented here. I'll be frank, I didn't believe it, I knew nothing about GPT3 at that stage.

Some More Interactions with GPT3

This post is just another annotated interaction designed to further illustrate how GPT3 is operating in preparation for the next post. 

Interacting with GPT3

It's been ages since I even looked at this blog, many more years since I was serious about sea ice studies. Given my nature there have been many interesting diversions, but overall, work has so consumed my energies that I have had little time or energy to blog. But over the last year I have found myself becoming more consumed with Artificial Intelligence and (what I consider to be) the ongoing Singularity.