Bibs & bobs #12

ChatGPT fun

Amid all the doom, panic and boosterism around ChatGPT there are tiny glimmers of fun. Here is one from Mark Schaefer: 20 Entertaining Uses of ChatGPT You Never Knew Were Possible. 

Accessible academic outputs

I mentioned the notion of embedding new text into LLMs which could then be interrogated. A model for this is Sahil Lavingia’s book, The minimalist entrepreneur. Here, you can ask his book questions. It’s roughly similar to Google’s talk to books. 

I began to wonder, what if academic publications all had an option like this. An option with which you could ask questions of the paper, chapter or book with the possibility of then extending your query via the references in the selected work. Yeah wild, but a fun shuffle of the academic game.

This would be a big step beyond what apps like Bearly offer currently. 

AI and writing

A collection of apps that use AI to support writing put together by Jeremy Caplan.

Transcription

There are a lot of options for transcripts of video and audio files. AssemblyAI is worth a play.  I was interested in a recent presentation by Venkatesh Rao to DEVCON in Bogota.  It mangled his name but the rest was pretty good. The app has a lot of options.

Formal Education and ChatGPT

As predictable as night follows day. Instead of doing the smart thing, i.e. given what machines can do some Oz universities will cling to pre-AI forms of assessment. So so so stupid.  The folk making these decisions are being paid big $. Who of them have explored or looked at how any of this is unfolding? Ban or domesticate is the old playbook formal education has used since the digital happened. It has not learned a thing from previous digital develops that go back to the late 1970s.  A measure of how smart any formal educational system or organisation is will be how well they deal with these developments, now and into the rapidly approaching future.  Below, I suggest the likely emergence of a new game that corporate universities will play: Whac-an-AI.

IMHO these developments will make Gutenberg seem like a tiny sneeze in the history of civilisation. 

Models, meddles and envy

The term physics envy is sometimes used to describe a motive behind some research in the sciences of the social. More broadly, the influence of different models drawn from science, i.e. Newtonian physics, chaos, non-linearity, complexity, emergent behaviour and so on, can be found in the logic that underpins a lot of research of the social and particularly in education which seems prone to picking up ideas that appear shiny or new, even if they are neither.

I’m of the view that there is nothing wrong with drawing on models and ideas from other fields but if you do, you need a decent helicopter view of the idea, its history and its limitations before drawing on it as metaphor or analogy.

For instance, if you wanted a quick and eloquent helicopter view of dimensions and the associated physics and mathematics it would be hard to go past a post of Margaret Wertheim’s. 

Perhaps it is an indication of how difficult it is to do good research in the so-called soft sciences that good helicopter views in this field of ideas, agendas and models are uncommon.  

Whac-an-AI

As each new bit of AI pops up to support writing, coding, planning and so on, and students studying in those fields continue to draw on them, you have a scenario for the perfect game of Whac-an-AI. In this game, the corporate university in order to protect it’s brand: our graduates don’t cheat, mistakenly commits to a practice of teaching their graduates how to do things that machines are now good at [1].

One thing that is predictable in this new wild space is that we will see more folk employed to manage and do the whacing. 

In anticipation of such a game emerging, I propose a new international standard for university stupidity called the Whac-an-AI.  Universities can be scored on an open ended scale by the number of moles they have more or less managed or tried to whac. Other awards spring to mind like best AI whac of the year, the most diligent AI whacer and so on. In time it could rival the ubiquitous ratings of universities that some universities appear to be obsessed with. If only :).

Sadly, if a tiny fraction of the effort associated with Whac-an-AI was put into:

1 supporting student understanding of what is going on in terms of LLMs and there many relatives

2 teaching students how to write half decent prompts and 

3 teaching students how to evaluate what LLMs output 

then life would be so much easier and better for students, staff and even managers!

The problem any university will face trying to ban or block AI use by students is that the moles will keep getting better and will breed rapidly. I am a bit of a fan of helicopter views of things. This tweet from Sterling Crispin puts the technical side of things into some perspective. 

Imagine an AI model that's 3x larger and more powerful than GPT3 aka ChatGPT

Google already built that in April, called PaLM, on their own TPU hardware competing with NVIDIA. People think ChatGPT will replace Google but they basically invented transformers in '17 (the T in GPT)

Imagine students playing with Monster moles!  The problem of course in an exponentially improving space is that the moles will keep getting better and will multiply faster than a corporate university could employ whacers.

Living through times that have elements that increase exponentially is an important part of making some sense of what is going on. As Dan Shipper recently wrote:

In his 1999 book The Age of Spiritual Machines, Ray Kurzweil wrote: “It is in the nature of exponential growth that events develop extremely slowly for extremely long periods of time, but as one glides through the knee of the curve, events erupt at an increasingly furious pace. And that is what we will experience as we enter the twenty-first century.”

A long time back I opted to use EdExEd as a label for my agenda. The Ex in the label is for exponentials. The other two: edges and education. Education should be plural I think.

                                                                                                    

[1] This is not a new phenomenon. Each new way of doing things has been met with attempts to ban or domesticate in formal education. A long time ago the hand held calculator was subject to banning and eventual domestication. Rarely was any thought given to the complementary knowledge and skills needed to make good use of the new way of doing things. There are lots of examples of this Dalek mindset.