Michael Piotrowski @true_mxp@qoto.org

If the first non-whitespace non-comment thing in an #Inform7 source file is a double-quoted string, the story title is set to that string. If, on the same line as the closing quote, there's a " by " then the story author is set to whatever follows it. The story author may be double-quoted, but doesn't have to be. (If it's single-quoted then literal double-quote characters will be part of the story author value.) There may be a period immediately after the story author.

If anyone is wondering if Bing+GPT is somehow going to be better than Google Bard, remember that both LLM systems will be dependent on the underlying search algorithms.

Here is Bing (prior to GPT integration) doing the wrong thing. Bing search is retrieving articles about how Google summary boxes are wrong and then giving the wrong answer.

Both will have the same failure modes. The only difference is that Google stepped in it first.

Source: https://twitter.com/stilgherrian/status/1623576572015050753

#ChatGPT

Mindshadow
Developed by Interplay
1984

Couldn't get a good source for the credits but it seems like it was designed by Rebecca Heineman and Allen Adham?

#DHNB2023 Calls for Reviewers for the Conference Proceedings! The DHNB proceedings are published with CEUR workshop series (https://ceur-ws.org/), a diamond open access publisher. Full papers are due 15 March 2023, and the review process will last until 26 April 2023.
Reviewing for the proceedings is an excellent way of contributing to the DHNB community and to digital humanities scholarship!
Interested? Fill in the short questionnaire, and we will get in touch! https://forms.gle/EEiaCgW56FWCHd4K7

Another new #Markdown extension in #pandoc 3: "mark".
pandoc --from=markdown+mark
This adds support for
==highlighted text==
syntax. Works with LaTeX, HTML, and docx output.
#formatExtension #highlight

Started work on SchXslt2, the second iteration of a modern XSLT-based ISO Schematron processor. SchXslt2 will be XSLT 3.0 only. What I can tell so far: The Schematron to XSLT transpiler is less then 500 lines of XSLT.

https://github.com/schxslt/SchXslt2-Core

#schematron #schxslt #dsdl #validation

I spend part of my academic life in the Digital Culture program at the University of Bergen. We’re hiring an associate professor dealing with game studies & digital narrative https://www.jobbnorge.no/en/available-jobs/job/240043/associate-professor-in-digital-culture

Post-Doc position at the @uzh Center for Reproducible Science: Participate in research projects related to reproducibility, Open Science and meta-science.

Position starts April 1st, 2023 or as soon as possible. Review of applications will start February 20, 2023 until the position is filled.

https://jobs.uzh.ch/offene-stellen/postdoctoral-position/9e669128-c162-44bf-b752-54f4b773df71

#academicjobs #OpenScience #postdoc #reproducibility

To celebrate the Kickstarter for Shift Happens going well, I thought I would show you 50 keyboards from my collection of really strange/esoteric/meaningful keyboards that I gathered over the years. (It might be the world’s strangest keyboard collection!)

When I first checked out Tcl in the early 1990s, interactively building GUIs with Tk on Unix felt magical. Back then the alternative was to batch compile endless, boilerplate-filled C sources to get a fraction of the functionality.

This interesting interview with Tcl's creator John Ousterhout puts these innovations into perspective by discussing the impact of the language:

https://pldb.com/posts/JohnOusterhout.html

#tcl

Are LLMs make it easier to write good essays? I’d say no, they are just shifting the goalpost.

LLMs will help us to write "correctly". However, crafting an original and engaging essay that someone wants to read will become a much more challenging task in the future.

https://www.technologyreview.com/2023/02/07/1067928/why-detecting-ai-generated-text-is-so-difficult-and-what-to-do-about-it/

@CerstinMahlow This is pretty useful:

https://towardsdatascience.com/the-carbon-footprint-of-chatgpt-66932314627d

@shriramk The numbers I see publicly claimed are 2-4 cents per chat. https://lifearchitect.ai/chatgpt/#cost

Scaled up to 8.5 billion searches per day, that's a bit of cash (and, energy).

I'm a little puzzled what the added value is, because the model optimizes word patterns, not facts. Sometimes the word patterns encode facts, sometimes, they do not.

@shriramk I only found this one

“Martin Bouchard, cofounder of Canadian data center company QScale, believes that, based on his reading of Microsoft and Google’s plans for search, adding generative AI to the process will require “at least four or five times more computing per search” at a minimum.”

here

https://www.wired.com/story/the-generative-ai-search-race-has-a-dirty-secret/

So one would have to know generic search costs to calculate it

Disruption and innovation in #search don’t come for free

The costs to train an #LLM are high

More importantly, inference costs far exceed training costs when deploying a model at any reasonable scale

Some Microsoft CEO Satya Nadella quotes:

o This new Bing will make Google come out and dance, and I want people to know that we made them dance

o From now on, the [gross margin] of search is going to drop forever

https://www.semianalysis.com/p/the-inference-cost-of-search-disruption

Wharton professor Ethan Mollick has an interesting piece on how easy it was for him to create an entirely-synthetic video of him lecturing, using AI text, voice, and video generation tools.

https://oneusefulthing.substack.com/p/a-quick-and-sobering-guide-to-cloning?utm_source=direct&r=i5f7&utm_campaign=post&utm_medium=web