One concept I likely will be adding to The Elements of Digital Ethics is ”Linguistic Misdirection”. Very much inspired by @emilymbender ’s objections to the use of AI.
Do listen to this episode of Tech Won’t Save Us, where Bender and Marx towards the end get into ”how the tech industry uses language to trick us into buying its narratives about technology”: https://techwontsave.us/episode/163_chatgpt_is_not_intelligent_w_emily_m_bender
”Wishful Mnemonic” is the term Bender uses in the show. Coined by Drew McDermott in 1976.
She gives the example of Automatic Speech Recognition (falsely attributes cognition) which should just be Automatic Transcription.
Some other examples of terms that may give an impression of something going on that really isn’t:
- Cognitive Computing
- Machine Learning
- Neural Network
Would love to hear some more examples of misleading terminology in the tech space in general, if you’ve got them. ”Cloud”, ”Smart” and ”Autonomous” would absolutely qualify.
#AIHype #DigitalEthics #DoubleSpeak #NewSpeak #WishfulMnemonic
Tagged: Akkoma
@jtrem
Here is a guide by @jamesmontemagno on how to accomplish this with IFTTT and Mastodon:
https://montemagno.com/automate-posting-to-mastodon-via-web-requests/amp/
I know I've seen other services that can be set up to do this as well.
#RSS2AP
Given all the controversy around Substack (again) I just want to promote the idea of self-hosting your own blog/newsletter using Ghost.
This allows you to set up your newsletter very much in the same way as Substack, with different tiers and a connection to Stripe. And it’s all yours, on your own domain, and without having to associate with bad actors.
While I myself don’t do paid tiers (I want all my content available openly), I have shared my way of setting up a VPS running Ghost within this blog post:
https://axbom.com/indieweb/
If I gauge enough interest I’m considering writing a post about only running a Substack alternative using Ghost..
#Indieweb #Substack
If a VPS is not your thing, and you want a really cheap option for just running a Ghost blog/newsletter with the possibility of paid subscription tiers, FastComet is the only hosting provider I know of that gives you a 1-click install of Ghost on a shared hosting solution.
Honestly, to get started in a day with a Ghost setup to replace Substack or any other newsletter/blog solutions I believe this could be the killer app. This currently starts at $2.19 per month.
https://www.fastcomet.com/ghost-hosting
#Indieweb #Substack
@Scmbradley
Inside Out, definitely.
So there is this film category, psychological thriller, that is supposedly different from just a thriller. But wouldn’t this suggest that we should also have psychological comedies and psychological romance films?
Anyway, that was my absurdist thought for the night. Wishing everyone a pleasant weekend.
Finally a well-versed walkthrough of how ChatGPT works that more people can take to heart. No we are not even close to something that "understands" our input. And maybe, just maybe, we can stop calling these things AI. But I won't hold my breath.
https://www.newyorker.com/science/annals-of-artificial-intelligence/what-kind-of-mind-does-chatgpt-have
#AIHype #DigitalEthics
Writing this book has been on my mind for two years now. By breaking it down into ~42 blog posts I think I can get it done in a reasonable amount of time.
Just getting started and will be scouting for publishers. Any words of advice are welcome.
Telling you all this so you can hold me accountable over the next 7 months. My goal is to finish writing by the end of the year. 😁
https://debook.axbom.com
#DigitalEthics #AIHype #ResponsibleInnovation
If you're curious about the content you can read a blog post of mine from March 2021 that covers all the elements (though a couple will be added for the book):
https://axbom.com/elements/
@dailygrail Haha, that would have been something.
@Homebrewandhacking @Mabande
We also must not forget her portrayal of Albert Nobbs.
https://slate.com/culture/2012/01/albert-nobbs-glenn-close-plays-a-woman-playing-a-man.html
I realised I should re-watch Hook to see if Robin Williams ever says “Captain my captain”. And then I started researching Hook online and now I really have to watch it again, because…
There are cameos by Carrie Fisher, Phil Collins and Glenn Close(!). Close plays one of the pirates. And I did not remember any of this. Also she is completely unrecognizable.
Photo:
https://legacy.throwbacks.com/content/images/2017/12/HookGutless.jpg
Haven't stopped thinking about this and may do a bit of an oldies marathon this weekend. Starting with Robin Williams and Glenn Close in The World According to Garp.
@kev
That is seriously cool. Now I have to start a separate blog just to try this 😅
@vinnatron
@LeoR1010 Tack, verkar ha gått bra. Jag vet inte om jag lyckades med uppdateringen för det är inget som ser annorlunda ut, men inget gick i alla fall sönder och det är ju alltid något att vara glad för.
Later today I’m doing a manual update of my SSL certificate for this server and I guess pretty much anything can happen.
I’ve had to do a fresh install of certbot since my first install of Akkoma so I have no idea if all the things will be added to all the right places.
Anyway, now you know in case my instance goes offline later today… 😅
That update was very smooth. Not even a glitch. Now to try an Akkoma update... my very first 😬
How much do you use and rely on emojis in different texts you write? Here is a great article on things to look out for related to meaning, display on different devices and services, low-vision and screen readers.
https://www.tiny.cloud/blog/emojis-and-accessibility/
#a11y #accessibility #emojis
@carl @anders
You should have been tagged in our continued conversation about domain knowledge Carl :)
One thing (of many things*) that makes me wary of using ChatGPT is because I don’t know how well I can judge my own autonomy within the use of the system. I feel like I am trapped within someone else’s imagination of what is important, rational or mathematically likely. And there is a clear risk of trusting quick responses in a desire to move along and get something ”done”.
If I for example would use it to trigger new plausible thoughts based on an input I would feel that I have skipped important steps that would either (a) allow me to see weaknesses in the new idea or (b) provide understanding so I can use the idea in the best possible way.
My favorite way of moving my thinking along is these types of conversations with you both and others who like to take their time to look at something from any and every perspective before moving forward too quickly. 😊
One area I truly am intrigued by is using these tools to write code. In my mind this could be very useful in moving power away from centralised and monocultural ways of manufacturing software.
*Other concerns of mine of course relate to ethics, energy-use, the expressed values of the organisations providing the tool, colonialist tendencies, power concentration, unaccountability and more. But I’ve written more about this elsewhere :) I am aware that this makes me biased to discredit. I have also realised that these new tools likely have created a need for me to add one or two elements to The Elements of Digital Ethics, related to machine-generated content: https://axbom.com/elements/
@anders
Yes, I’m also critical of all the familiarity being built into these designs (such as using ”I”, ”me” and emojis). Or the certainty with which responses are expressed.
@anders This I certainly agree on. I had this in the back of my mind as well as I was writing. I have for example saved all my emails since 1998. And lots of other writing of course as well, including my blog. Being able to source this material using a language model could provide useful. Almost like being able to retrieve data from a diary to make note of shifts in ideas, culture and values over time. So within a limited scope and source material that I am failiar with it makes a lot of sense.
@anders
This is fair, and relatable to telescopes. I still need domain knowledge to understand what it is I am seeing through my telescope. And this is also what give me a healthy amount of skepticism.
And I think this is key to these tools, how important it is to have a certain domain knowledge before using it.
One specific problem could then be described as putting this tool in everyone’s hands without the domain knowledge to understand or judge what they are seeing in the other end of it.
@anders I think my primary tools for augmenting my own thinking are sketching, note-taking and writing (and some tools similar to what you describe that can help me get an overview of my own notes and sketches). And in some sense reading books. I believe my inability to trust the output of ChatGPT, and the lack of transparency on what it is trained on, makes it difficult for me to find a good use for it. 😬
I do understand the idea of using it for critical thinking and generating statements that are to be judged and reasoned around, as the following post explains, but I rarely see this type of critical thinking even in examples provided by peers in my industry – using it in workplaces. The amount of trust many people place in the output, and use as conclusions for analytical work, is a problem…
https://www.cambridge.org/elt/blog/2023/03/30/enhancing-learners-critical-thinking-skills-with-ai-assisted-technology/
When we augment with a telescope it is because we have recognised a need to see further into space. When we augment with a hearing aid or speaker, it is because we have recognised a need to hear better. And we can immediately validate if they work for this purpose.
It is not entirely clear to me what part of human thinking is augmented with ChatGPT and how I can judge if that is actually what is happening. i.e. I woud like to see the need more clearly described, and I would like to see the result of that need being met more clearly validated. Can everyone think better with a tool like ChatGPT, in the same way everyone with vision can see further with a telescope?
For me a microscope and a hearing aid can eliminate misunderstanding, as they help me see or hear more clearly. Meanwhile a tool like ChatGPT can *boost* misunderstanding as it is a statistical representation of a source that is unclear to me, and evidentially biased.
I know you didn’t say ChatGPT itself was augmenting thinking, as you used the word catayst. But I did find it interesting to think of it as an augmentation tool, where ChatGPT clearly can be used to augment all the wrong things (disinformation, bias, depression, etc).