Limits of ChatGPT

With the world discussing AI that writes, a recent post from Bret Devereaux at A Collection of Unmitigated Pedantry offers a useful corrective, both about how present-day large language models like GPT-3 and ChatGPT are far less intelligent and capable than naive users assume, and how they pose less of a challenge than feared to writing.

I would say the key point to take away is remembering that these systems are just a blender that mixes and matches words based on probability. They cannot understand the simplest thing, and so their output will never be authoritative or credible without manual human checking. As mix-and-matchers they can also never be original — only capable of emulating what is common in what they have already seen.

Bugs in GaiaGPS

They are too busy/understaffed to communicate with anyone or fix any bugs, but in case anyone at GaiaGPS ever has some free time, these are issues I have noticed.

Some may be on account of the large number of tracks which I have collected.

  1. On both my phone and computer, the map at gaiagps.com or on the app will not show all of my tracks when zoomed out. Far enough out (like looking at the whole city or a big part of it) they do not show at all, and then they pop into vision when I move in enough
  2. Track names often fail to sync from my iOS phone to the web/desktop app. They show up as generic track names and I need to figure out which is which and manually rename them
  3. Both the desktop and especially the iOS version insist on showing waypoints near tracks which are visible, even when the folder for those waypoints is explicitly set to not display
  4. When exporting a folder of tracks with waypoints near them, GaiaGPS includes some of the waypoints along with the tracks in the exported .GPX or .KML file
  5. Tracks and waypoints often fail to move into folders, on both desktop and iOS. I select a track or some waypoints and put it in a folder. It may disappear from where it was previously, but if I wait a minute it will often jump back to where it was before.

None of these break the functionality of the app, which is remarkably capable for a free version. Still, it would be nice to see a good product improved by sorting them out.

Mapping and cartography

When the University of Toronto gyms closed for the pandemic, I realized and decided that I would be getting most of my exercise through walking.

All through writing up my dissertation, I broke down my life into 20 minute chunks of either thesis work, sleeping, exercising (walking), or doing anything else.

I am now up to well over 7,000 km of walking since August 2020. This had begun to produce spin-off projects: the Toronto Urban Hike Collection to identify nice nature trails, maps made in QGIS and posted to Flickr, and animation and wall-sized map projects that I am still developing.

The soundest base for a diagnosis

So much depends on style, that factor of which we are growing more and more suspicious, that although the tendency of criticism is to explain a writer either in terms of his sexual experience or his economic background, I still believe technique remains the soundest base for a diagnosis, that it should be possible to learn as much about an author’s income and sex-life from one paragraph of his writing as from his cheque stubs and his love-letters, and one should also be able to learn how well he writes, and what are his influences. Critics who ignore style are liable to lump good and bad writers together in support of pre-conceived theories.

Connoly, Cyril. Enemies of Promise. George Routledge & Sons, Ltd. Broadway House: 68–74 Carter Lane, E.C. 1938

Nobody to write for but myself

It has occurred to me that, while I am waiting to hear back on numerous job applications, and while I am waiting to graduate in absentia on March 10, I can shamelessly use the University of Toronto libraries to research absolutely anything of interest. Today, I got some guidance in digital cartography, was invited into a 2.5 hour workshop on making honest and educational infographics, and began reading Cyril Connoly’s 1938 Enemies of Promise in search of backstory and inspiration for my Sherlock Holmes pastiche project.

By the way, while it hasn’t always been treated with perfect gentleness, and the repairs are such as would be used for a practical object rather than a museum piece (bent corners and pencil-marks abound inside, and whoever taped the barcode on to the cover didn’t think tearing the tape unevenly would mar its appearance), based on the front matter I think the Connoly is a first edition.

I intended the title, incidentally, to refer to self-motivated writing not being actively overseen, edited, or hurried along and not to suggest that nobody is reading my writing or that I don’t care about those who do. In fact, I am always curious to know who is still lurking around here after all these years, or who has stopped by because they found a single post of interest.