Prompt

Describe or narrate a scenario about a beverage found a decade into a future in which a profound historical evolution has occurred. Your description should address issues related to artificial intelligence and elicit feelings of hope.

I set a timer for 4 minutes initially, then did another 4 minutes for my initial writing just because I was enjoying myself. From there I generated some add-ons using Suno.AI, Adobe Firefly, and Microsoft Copilot and then reflected on the process.

Response

This beverage is completely personalized, as your self-hosted AI assistant has access to exhaustive data about your preferences, needs, and sodium levels. It’s different depending on your context and comes from a self-managing reusable container that you carry with you. After the proliferation of AI assistants, open source alternatives have saturated the market meaning you can self-host and manage your own AI assistant, comfortable that your data isn’t being sold to a third party, and that your assistant prioritizes you and your needs above the needs of advertisers.

The beverage is made from raw materials that are shipped to individual households. These raw materials are used by the AutoChef to either create meals or your own ingredients if you prefer to make your own food. Just like AI assistants, AutoChef used to be a brand but has since genericized to be a catch-all label for the hacked together and self-hosted variety of AI enabled material processing systems that exist on an individual household or small community level.

While there was efficiency gains to be had by the centralized cloud hosting model of the internet in the early 2020s, it was determined that the rebound effect meant that companies were gobbling up any efficiency gains in an attempt to analyze more and more data to better understand their audiences. Consumers rejected that model after waves of cloud hosted service outages and have returned to a more fractured self-hosting model for many day-to-day services. It’s less efficient but the overall use of compute has dropped globally due to the many overall lifestyle changes that came about because of the disappearance of artificially cheapened cloud computing.

This overall trade off – less efficiency in exchange for less overall compute has been reflected by all facets of society – humanity does less, works less, and consumes less.

AI Brew Artifacts

Reflection

I was inspired mostly by Smyth et al. (2021) and their description of future past speculations – specifically Futurama. Visions of the future designed specifically around products and not people make my head spin. Harari (2017) speaks of possibilities enabled by the AI revolution. I’m tired of reading how the AI revolution will enable humanity to focus on more creative forms of work – did that happen with the spinning jenny? Did we find ourselves luxuriating in the countryside, reclining in creative idleness? The universal basic income (UBI) project in Ontario that Harari references has already been spun down (Ontario, 2018) even though respondents “consistently reported improvements in their health, housing situation, financial status, family relations, and labour market experiences” (Ferdosi et al., 2020, p.6). 95.5% of respondents reported the cancellation of this pilot forced them to abandon or place on hold their life plans (Ferdosi et al., 2020, p.60). Federosi et al.’s (2020) reporting is limited because the team that was created to evaluate the impact of the UBI pilot was disbanded at the same time the pilot was prematurely ended.

I am exhausted of futures envisioned around the structure of consumption, and growth. The most hopeful message that I can think of now is a world of less.

References

Harari, Y. N. (2017). Reboot for the AI revolution. Nature (London), 550(7676), 324-327. https://doi.org/10.1038/550324a

Ferdosi, M., McDowell, T., Lewchuk, W., & Ross, S. (2020). Southern Ontario’s basic income experience. McMaster University. https://labourstudies.socsci.mcmaster.ca/documents/southern-ontarios-basic-income-experience.pdf

Ontario. (August 31, 2018). Ontario’s government for the people announces compassionate wind down of basic income research project. Ontario Newsroom. https://news.ontario.ca/en/release/49980/ontarios-government-for-the-people-announces-compassionate-wind-down-of-basic-income-research-project

Smyth, M., Auger, J., & Helgason, I. (2021). Echoes of futures past – Speculations and fictions from history. In I. Mitroic, J. Auger, J. Hanna, I. Helgason (Eds.), Beyond speculative design: Past present future (pp.24- 67). SpeculativeEdu; Arts Academy, University of Split.

This feels right in my wheelhouse, to be honest. I was a web designer and developer for ten years – lovingly crafting dark, medium, and light patterns of my own. The Golden Rule as I saw it was:

1. Make it easiest for the user to do what you want them to do

This all hinges on you knowing what you want the user to do and being effective in the interventions you make to guide them through your ideal process with little time, effort, or thought required on their part.

This game really was a delight – every trick, every deceitful button, every pattern that has ever bothered me felt like it was represented. Alphabetical order month drop-down, check. Age slider, check. Select/unselect all buttons buried in a list, check. My personal favorite, and one that I see contravened in genuine interfaces all across the big beautiful web is: the button on the right doesn’t go ‘forward’ in the process. Check!

Thinking about advertising and these dark patterns got me thinking about Zalewski’s 2009 US Patent US8246454B2 or the infamous “say McDonalds to end commercial” patent.

The “embodiment” (Zalewski, 2009) of advertising in this patent takes things a step further in that it demands a physical interaction, instead of racing “to the bottom of the brain stem” (Harris, 2017) to influence thoughts before they would become movement. I wonder if the reason that we haven’t seen as much embodied advertising is because the techniques that Harris addresses in his TED talk are just that much more effective.

Harris’ (2017) note of “technology is not neutral” was a standout for me from the TED Talk (Is it ironic that the video of his talk auto-played when I visited the TED site looking for citation information as, in another tab, he talked about market share and the effectiveness of auto-playing videos?). His view of a Facebook that guides users into the best use of their time reflects his own values and circumstances in a distinctly non-neutral way. His example of dinner at home with friends to have controversial conversations is only possible when you live in a location where others can physically attend, when you have the time, energy, and funds to host people, and when you have friends that you can have controversial conversations with. His own vision of best-case scenario completely disregards the advantages of an online platform itself – low cost, asynchronous, geographically diverse connections. I’m not sure that I have a point beyond both agreeing and disagreeing with Harris – technology is not neutral, and neither is your own vision of a best-case scenario.

Tufecki’s (2017) TED Talk also engages with advertising interventions happening and working startlingly well for reasons we don’t fully understand. I’ve honestly thought for some time that when speakers say things like “we don’t fully understand” why machine learning algorithms arrive at certain conclusions, they were really saying “I don’t fully understand”. I’m realizing with creeping horror that maybe “we” actually don’t understand fully. For example, I have a difficult, if not impossible, time reconciling the idea that I could deploy a chatbot directly to students that would effectively sum up lectures and provide individualized tutoring to many students at the cost of some who are led on a wild goose chase by hallucinations. At the same time, here we are, living in a world that increasingly appears to be based on just that sort of model where we’ve bet it all on vaguely understood technologies that work great for reasons unknown.

References

Harris, T. (April 2017). How a handful of tech companies control billions of minds every day. TED Talk. https://www.ted.com/talks/tristan_harris_how_a_handful_of_tech_companies_control_billions_of_minds_every_day?subtitle=en

Tufekci, Z. (2017). We’re building a dystopia just to make people click on ads. TED Talk. https://www.ted.com/talks/zeynep_tufekci_we_re_building_a_dystopia_just_to_make_people_click_on_ads?subtitle=en

Zalewski, G. (2009). System for converting television commercials into interactive networked video games (US Patent No. 8246454B2). US Patent and Trademark Office. https://patents.google.com/patent/US8246454B2/en

This song selection visualization shows our selections as an unweighted network. “Within networks, how connected an individual node is becomes a key metric of its significance within the network”, so how commonly selected a Golden Record song could be interpreted as that song’s significance (Systems Innovation, 2015). However, I think this type of data visualization does not articulate qualitative selection factors. For example, while we were all ultimately curating a selection of the songs on the Golden Record, there was no unifying approach to how we ranked them so it’s difficult for me to determine just what conclusions we could raise from this data set.    

Before thinking more about drawing conclusions, I thought of investigating the data. It appears that Shannon Wong, Stephanie Takeda, Abdulehed Yarkin and I each had five selections in common so their criteria compared to mine might provide an interesting vector to approach this data from.  

I looked at Shannon’s webspace and saw that we did not use a similar sorting mechanism to curate our lists as they worked off geographical representation as a sorting structure (Wong, 2024). Stephanie and I came at this task completely at odds – they note “my decisions were primarily based on criteria other than personal preference” which is at direct opposition with how I just selected based on my own preference (Takeda, 2024). Abdulehed and I were on a somewhat similar approach– as part of their sorting algorithm Abdulehed ranked pieces with higher emotional impact as higher (Yarkin, 2024). While I didn’t take that exact approach there’s more common ground there than with Stephanie and Shannon’s approaches. 

After exploring the posts of those I had a lot in common with I wondered if I could filter the visualization to show me who I didn’t have anything in common with. Filtering by “source” gave me another vector to play with the data. It appears I have a selection in common with everyone in the class, but Tina Wei and I are the least networked – only “Dark was the Night” joins our curated lists together. What I found funny about this is that “Dark was the Night” is my far away favorite track on the list!

I read Tina’s post about selection criteria and our algorithm did have much more in common than the criteria of the colleagues that I had more selections in common with. Tina and I both leaned into the subjective and made selections without a logically consistent selection framework (Wei, 2024).  

I find myself again thinking about Linda Salzman Sagan’s reference to the “ridiculous and the sublime” when selecting songs for the Golden Record (Taylor, 2019). Those who I have selections in common with, I don’t have criteria in common with. Those who I don’t have selections in common with, I do have criteria in common with! It’s getting muddier and muddier for me to try and draw conclusions from this data.

Looking for another way through, I filtered to find the most popular tracks. Melancholy Blues and Johnny B. Goode are the most selected songs with 16 students selecting them. Does that reflect the dominant mode of cultural production in our contexts, or personal taste, or are these just objectively the best songs to represent humanity to a new form of life?

Re-listening to Melancholy Blues (ugh, maybe I should have selected it) and scrolling through the YouTube comments I’m shocked that they are largely positive and happy, anticipating a new audience for Louis Armstrong, assured of the positive reception of this song, excited that Louis Armstrong’s work and output get to live on in space (Lunathicka, 2010). Another possible motivating factor for song selection enters the fray – not possible reception of the song, but some occult math that could be used to calculate how deserving the artist is of their work being preserved and shot into space. 

As an aside – I was driving back from Kelowna last weekend, baking in the heat, and talking to my husband about the original Golden Record task. We talked through what we would do if we had to do the Golden Record from scratch all over again. He’s adamant that the best approach would be to take most chart-topping hits from the last X number of years. I’m aghast at his suggestion, that what we can afford to lose is the already underrepresented voices, he argues that the point of the task is already accomplished by song popularity. Ultimately, we agree that the task leans into the absurd and that we’ll just have to approach it differently if either of us is called upon to curate Golden Record 2.  

So, after much looking around, poking and prodding, talking and arguing, I return to the original questions of the prompt. This visualization cannot capture why selections are made in this context; I think partially due to a lack of common decision-making algorithms. I believe there is an opportunity to collect additional data around the songs that were emphatically not selected – an inversion of the selection task where you eliminate songs may be an interesting way to make a contrasting data set from which you could draw conclusions around which songs are controversial or polarizing.  This type of visualization helps to articulate the quantitative but does not involve the qualitative which on a deeply individual task like Golden Record curation means the missing data has a deep impact on the conclusions that can be drawn.

References 

• Lunathicka. (2010). Voyager’s golden record: Melancholy blues_Louis Armstrong and his hot seven {Video]. YouTube. https://www.youtube.com/watch?v=T9BC5PY_OaM 
• Systems Innovation. (2015). Network connections [Video]. YouTube. https://www.youtube.com/watch?v=2iViaEAytxw&t=48s  
• Takeda, S. (2024). Task 8: Golden record curation. ETEC 540 Webspace. https://blogs.ubc.ca/stephtakeda/2024/07/07/task-8-golden-record-curation/  
• Taylor, Dallas. (Host). (April 2019). Voyager Golden Record (No. 64) [Audio podcast episode].  In Twenty Thousand Hertz. Defacto Sound.  https://www.20k.org/episodes/voyagergoldenrecord 
• Wei, T. (2024). Task 8. Wei to Education. https://blogs.ubc.ca/thisisthewei/task-8/ 
• Wong, S. (July 2, 2024). Task 8: Golden record curation assignment. Shannon’s Blog. https://blogs.ubc.ca/shannonwong/2024/07/02/task-8-golden-record-curation-assignment/ 
• Yarkin, A. (July 5, 2024). Task 8: Golden record curation. Abdulehed’s ETEC 540. https://blogs.ubc.ca/akyar540/2024/07/05/task-8-golden-record-curation/ 

This was the most difficult assignment for me, so far.

I really appreciated Linda Salzman Sagan’s reflection on the process of collecting the sounds for The Sounds of Earth: “there was a sense of wonder to it and a sense of the ridiculous and the sublime” (Taylor, 2019). At the end of the day music is not something I have any training in, so I can’t come from the perspective of evaluating the kinds of music or what entries in musical history are particularly worthwhile to preserve. I thought about types of sound, countries of origin, population centers as all possible ways to sort and prioritize these tracks.

However, this is such a small selection that it demands the perspective of thinking about “what can we afford to lose” as mentioned by Dr. Smith Rumsey (2018). Practically speaking, if we are curating all the noise that people make to ten entries, we are losing everything. So having said all of this, and leaning into the sense of the ridiculous, these are just the ten tracks that I enjoyed the most while listening to this record.

Track list – titles taken from the NASA Golden Record track listing

• “Dark Was the Night,” written and performed by Blind Willie Johnson. 3:15
• Senegal, percussion, recorded by Charles Duvelle. 2:08
• Australia, Aborigine songs, “Morning Star” and “Devil Bird,” recorded by Sandra LeBrun Holmes. 1:26
• Japan, shakuhachi, “Tsuru No Sugomori” (“Crane’s Nest,”) performed by Goro Yamaguchi. 4:51
• Navajo Indians, Night Chant, recorded by Willard Rhodes. 0:57
• Beethoven, String Quartet No. 13 in B flat, Opus 130, Cavatina, performed by Budapest String Quartet. 6:37
• Solomon Islands, panpipes, collected by the Solomon Islands Broadcasting Service. 1:12
• Azerbaijan S.S.R., bagpipes, recorded by Radio Moscow. 2:30
• Georgian S.S.R., chorus, “Tchakrulo,” collected by Radio Moscow. 2:18
• China, ch’in, “Flowing Streams,” performed by Kuan P’ing-hu. 7:37

References

• Brown University. (2018). Abby Smith Rumsey: “Digital Memory: What Can We Afford to Lose?” [Video]. YouTube. https://www.youtube.com/watch?v=FBrahqg9ZMc
• Taylor, Dallas. (Host). (April 2019). Voyager Golden Record (No. 64) [Audio podcast episode].  In Twenty Thousand Hertz. Defacto Sound.  https://www.20k.org/episodes/voyagergoldenrecord

This week I thought about what modalities included audio, could engage with the photos from the “What’s in your bag?” task, and would also move beyond just spoken word audio to include other modalities.

Ultimately I decided on creating a short podcast episode. This podcast is hosted by an AGI system and takes place in the “near distant” future. This AGI system is a bit of a humanity aficionado, and the show focuses on different ways humanity would exist in the world. In this episode the AGI is able to secure a permit to thaw an entry from the library of humanity to talk about how humanity would physically exist in the world and move from place to place.

I was thinking about the great multimodal diagram from Cazden et al. (1996) so wanted to create a visual, audio, and linguistic artifact. To that end I used a variety of tools and services and within the fiction of the episode I wanted to unadulterate the products of the systems as much as possible.

• Copilot to create the podcast logo/image
  • I did try a couple of times to get it away from such stereotypical AGI imagery and incorporate more elements of podcast splash screens/logos. Eventually I thought, maybe my ideas around what makes a good podcast image are limited by my frail humanity so I left it as-is.
• Suno.ai to create the podcast intro/outro music
  • I’m not really a music person, so this I left as-is without any micro-managing. The prompt was just around creating podcast theme music for a podcast hosted by an AGI in the near distant future.
• ChatGPT 3.5 to create the script for the AGI host
  • Again, I kept this simple and used the first output. It’s more long-winded than the script I had drafted but I realized I wanted to adhere to the in-fiction rules as much as possible.
• FreeTTS.com to generate the audio of the host from my script
  • This is far from the best AI generated voice I’ve heard but is listenable!
• Kaltura’s machine-generated captions to create captions of the final product

Finally, I wrote myself a script for my section, recorded myself, and edited it all together. I’m hopeful that this final product ticked the same boxes the images did, in a playful way.

References

• Cazden, C., Cope, B., Fairclough, N., Gee, J., & New London Group. (1996). A pedagogy of multiliteracies: Designing social futures. Harvard Educational Review, 66(1), 60-92. https://doi.org/10.17763/haer.66.1.17370n67v22j160u

WordPress seems determined to embed my emojis as images which breaks my careful formatting around line-breaks so behold: a screenshot.

Reflection

Working on this emoji story I selected my story largely based on what media I had consumed recently, eliminating a more recently consumed option that wouldn’t have been fun or entertaining for anyone, myself included due to how I didn’t even know where to start. A big thank you to my friend for making us all watch Zardoz (1974) last night, where would I even begin. So, minor spoilers, this isn’t Zardoz (1974).

As I was developing my emoji story I spent some time thinking back to my undergrad in fine art. The selection and placement of emojis reminds of a formal analysis of ancient Egyptian art when compared to more naturalist approaches of Greek and Roman. In ancient Egyptian art this person is larger because they are more important, they are portrayed ahead of the others because they are leaders, there is often text presented with images… not to even mention hieroglyphics and how we may view that as a blurring of text and image.

Admittedly, I use a lot of emojis in my day-to-day. I routinely find myself baffled by the emoji selection in MS Teams and it was a very good day for me when they allowed reacting with any emoji rather than the pre-set corporate starter pack of thumbsup, heart, etc. I found this entertaining!

Thinking about Kress (2005) specifically while composing this I found myself wondering about the immediacy of image compared to the unfolding of text. The emoji story finds a pretty comfortable middle ground. Kress (2005) also mentions considering both the fit of modality and content, but also the fit of modality and audience. I wonder about how my colleagues will find this task, how this course has provided a huge range of activities for assignments (tempered with writing for each one).

I also found myself reflecting on how this text is almost 20 years old – how have our ideas around multimodality changed in that intervening time? I wonder if we are now experiencing a similar semiotic revolution around video with the meteoric rise of short form video content in the last five years. It’s an interesting tension – video takes the audience back to the unfolding of written text, compared to the immediacy of the image.

References

• Kress, G. (2005). Gains and losses: New forms of texts, knowledge, and learning. Computers and Composition, 22(1), 5-22. https://doi.org/10.1016/j.compcom.2004.12.004

Files

blogs.ubc.ca appears to have html and zip file upload turned off so I’ve uploaded to Google Drive and shared below. My reflection will contain spoilers for my incredibly fascinating work so you might want to view the Twine first 🙂

HTML instructions to view the Twine:

• download the file
• navigate to it in your file explorer
• open it in a browser

HTML file: https://drive.google.com/file/d/1Ei8NpRQCou0r7hgPwAGPYJx-b6mQv-mN/view?usp=drive_link

ZIP instructions to view the Twine:

• download the file
• navigate to it in your file explorer
• open/unpack the zip
• open the .html file in a browser

ZIP file: https://drive.google.com/file/d/11XtY5mDXwlu_33QjB0cAANsdCWqMvtSw/view?usp=drive_link

Reflection

I truly have an enthusiasm for Twine as it turns out. Initially I was struggling to find an approach that resonated with me and bridged the gap between this week’s readings and my expectations of Twine. Ultimately I went with what I know best and started with a dark, stormy night and a mysterious lighthouse.

I bounced around a few ideas for mapping this activity out before starting – in the end I was torn between Coggle and handwriting a map and the fact that I’ve been struggling with headaches this week settled the deal – anything for a little less screen time.

Going into the readings this week I was mostly considering ideas of information hierarchy in the context of my experiences as a web developer. Broken links were (and are!) a nightmare of lost SEO and frustrated users. Thinking about Engelbart’s conceptual framework (1963) helped me to broaden the horizons of my thinking beyond digital information structures to include the physical structures and training that people rely on when creating work. As I read Bush (1945) I was shocked at the prescience of his writing. Lots to consider around different types of input, data, and flexibility of information – and it’s in step with conversations I’m having today around multi-modal language models like OpenAI’s 4o.

I hope I was able to effectively bring the readings and ideas together with what I think is a story in step with the fine traditions of choose-your-own-adventure.

References

• Bolter, J.D. (2001a). The electronic book. In Writing space: Computers, hypertext, and the remediation of print (pp.69-85). Lawrence Erlbaum Associates.
• Bolter, J.D. (2001b). Hypertext and the remediation of print. In Writing space: Computers, hypertext, and the remediation of print (pp.29-45). Lawrence Erlbaum Associates.
• Bush, V. (1945). As we may think. The Atlantic Monthly, 176(1). 101-108.
• Engelbart, D. (1963). A conceptual framework for the augmentation of the man’s intellect. In P. W. Hawerton & D. C. Weeks (Eds.), Vistas in information handling, Volume I: The augmentation of man’s intellect by machine. Spartan Books.
• Nelson, T. (1999). Xanalogical structure, needed now more than ever: Parallel documents, deep links to content, deep versioning and deep re-use. ACM Computing Surveys, 31(4).

Images of Hand Writing

Reflection

I elected to write by hand for this text, because I knew I’d be typing this reflection and wanted to print, write, and type in my response to this task. I found this surprisingly difficult! 500 words is a really achievable amount for me to type so I went in very overconfident but found hand-writing and maintaining a train of thought challenging across that length.

While writing I mused about the procedural differences between typing and writing. Typing lends itself to the scattered, iterative, looping writing process that I’ve developed. When I hand write it’s typically either in personal letters or my weekly planner so my editing process is more focused on adding stickers and washi tape and less focused on grammatical changes or editing for brevity and tone. I believe the choice to write really impacted how I approached the task – I felt much more pressure to have a clear idea of the final product as I worked.

An interesting side effect of writing letters that contributed to my writing process is that now I tend to outline a letter in a journal before writing it. This came about because I would send a letter and have no idea what I wrote in it – with no access to a sent folder I couldn’t remember what I had shared already! I think that dovetails with the ideas explored in both “Stuff to blow your mind” episodes around the impact of writing on the sharing or duplication of information and record keeping (Lamb & McCormick, 2020 May 26).

While my preference is informed by situation and contextually I do enjoy hand writing, it would be dishonest to ignore the idea that >90% of my writing takes the form of typing. My process, and my output, have been deeply impacted by the flexibility that typing and word processing software brings to writing tasks.

Thinking about formats for consuming information like they discuss in the second part of the invention of the book series – I don’t know that I could read a full book made from potato stamped letters, haha, but I do experience text in a variety of ways! For fun reading I typically reach for either a paper book or my Kobo, but for work I lean towards the laptop. The hosts of “Stuff to blow your mind” talk about reading books on their phones which I find mind blowing. I appreciate the opportunity take the time to consider more deeply how the format of text effects not just creation but consumption of text.

Potato Print

Reflection

I enjoyed the potato stamp process! I don’t have children but I do have a weekly Art Night where my friends gather to work on creative projects of all kinds and this was right over the plate for us. You can see the second image in the gallery shows the work my friends created. Because I integrated this into Art Night the process took about an hour for me with lots of chatting and iterative testing of the stamps as we worked. It was interesting to note that one friend arrived organically to the idea of multi color printing “if only I could do a mid tone” – you can through a second stamp! He was not invested enough to pursue a second stamp after his first one, but that was an interesting process to arrive at.

I selected the word “muzak” to print. I had attended a panel that day around generative AI use principles and guidelines at UBC and I found it interesting they lingered over the idea of “original thought”. I wrote on that in my handwriting a little bit, but I find the discussion around original thought quite interesting. In print-making particular – what is original when dealing with a series of reproductions? In thought – what is remix, what is constitutes originality? It smacks of goalpost moving, for me, the idea that there’s an inherent nugget of originality that only human intelligence can achieve… So why “muzak” instead of “music”? I was framing the elevator, easy listening, light, consumable approach of “muzak” compared to “music” as a parallel of generated text versus human written text. I think the space originality is interesting to explore as we are asking more foundational questions about the use of AI.

One aspect of the letters selected in my word “muzak” is that I do write z differently but couldn’t make the stamp work – something about the curve in my hand-writing of z was hard to capture and it wound up looking like a backwards 3 each time! Ultimately I chose to abandon my hand-writing approach and move to a different rendering of z for readability.

Generated Transcription of Hand Written Text

For fun, I thought I’d upload the images of my handwriting to try and automatically generate a transcript or image description of my written work. This was generated with Claude on June 6, 2024 and I have not edited it for accuracy as I’m interested in how the generated writing differs from the hand written text. Honestly, there are not that many mistakes considering the contrast of the pen I selected and my chicken-scratch!

4.4 Task 4: Manual Scripts

Overall, this task contrasts manual and mechanized forms of writing. I wonder about more fringe spaces like the use of AI-supported tools for writing such as Grammarly. Typing versus writing are different ways to approach text development and it impacts the final product in ways from start to finish. What about the layers of additional tweaks? I always get off-ice- suggestions to support brevity and conciseness in my writing, but when hand-writing I’m not getting those direct suggestions. As an exercise, I think I’ll go back and edit this – but when I’m writing letters or in my weekly planner, I don’t actively participate in that editing process in the same way as when I write for work or school using a computer… Circling back to AI – only after the AI course last term have I done really any editing using ChatGPT – overall I still haven’t incorporated it into my own workflows but I know many others have. From my perspective, it’s interesting to consider text as both output and input in these platforms, which is a real departure to how I naturally or currently consider handwritten text. I also typically approach writing in process differently in computer- or mechanized writing –

Image 2: For example I typically start with a structure and I write within that structure. As part of editing, I will move paragraphs around, to experiment with the flow. To contrast, when I am hand writing I feel more pressure to not require that level of foundational editing, as it would be more labour intensive.
This idea of labour is an interesting one to explore especially in this AI-impacted context. I’m in currently AI-operated writing can extend my mechanized writing approach with prompting. I could generate an outline and structure, then generate text, revise, restructure and re order it. How different is that from my mechanized writing which so easily enables me to make those edits and revisions? I’ve asked a few LLMs about human versus AI generated content and originality, and have found pretty unremarkable answers. Claude, which launched today in Canada, indicates that only human original creativity is created on a foundation of understanding of concept when compared to generating text based on probabilities, predictions.

Image 3: I find that unsatisfying / I’m sure I will continue to find many philosophical questions unsatisfying, as we continue down this AI-enabled road. What can I do to understand the gaps between A.I generated content and human generated content on a philosophical or epistemological level? Does the fact that I’ve slowly and painfully writtn this out versus typing it quickly and easily impacted this text? I bet it’s less cohesive than if I didn’t have to stop repeatedly when my arm starts cramping, such as my clutch on this pen. I bet if I typed this I could have been finished an age ago. What kind of point was I trying to make?

References

Lamb, R., & McCormick, J. (Hosts). (2020, May 26). From the vault: Invention of the book, part 1. [Audio podcast episode]. In Stuff to blow your mind. iHeart Radio.

Lamb, R., & McCormick, J. (Hosts). (2020, May 28). From the vault: Invention of the book, part 2. [Audio podcast episode]. In Stuff to blow your mind. iHeart Radio. 

Voice to Text using Windows

OK so I feel like the first thing that I have to say is that this is really uncomfortable for me that was something I really had to get over at the beginning of the pandemic was how uncomfortable and awkward and embarrassed I feel when I talk to a computer so that was like something that I I had to deal with I especially with zoom and when there’s technical issues still it it just like I don’t know I I feel terrible about it so I I feel embarrassed during this that’s my first thing the second thing I guess that I’m going to talk about is movie that I just watched so I finally watched dream scenario I am a pretty big Nicolas Cage fan so so I’m you know try to watch everything that he puts out and I think lately there’s been a cage renaissance of sorts but dream scenario was amiss for me which is interesting I think because on paper there’s a lot there that I should like I I would argue it was miscategorized by almost everything like the YouTube algorithm says that it’s a comedy we rented it off of YouTube I find that very interesting because it’s certainly not funny but at the same time I think it’s funny and maybe the same way that the office is funny which I can’t watch also because I find it so uncomfortable so maybe what I’m arguing is that the office is not a comedy either and so maybe I think it’s not a horror movie either although it seems like the marketing kind of leaned into the horror aspect and there is like a nod to that idea that it’s a horror movie but it but it doesn’t it does remain in that space enough I think to be fully called over film I I think I have to argue that it it’s creating a new genre of of of a new sub genre of horror maybe instead of something like body horror you know a horror that really focuses on embodiment of of like fear and and modifications and changes to your physical body I argue that dream scenario is maybe the first entry social horror genre and not one where you are looking at society as a horrifying thing but the social death of a of a person I guess is maybe what we’re talking about like watching someone who’s totally graceless just fail over and over and over again is social horror I think that’s I think that’s horrifying I would hate to for that to happen to me add but why I thought it was a miss I mean I think that it didn’t go very far in any direction and I think that is a complaint that I wind up happening with a lot of Nicolas Cage movies I wonder if it’s because I think that everything that he’s in should be so bombastic that when something is simply just I don’t know typical maybe I think that’s disappointing when it could be going to absolute extremes like to contrast this let’s look at at another couple of semi recent I say semi recent because time lately has been really compressed but semi recent cage films like you did the color out of space and that was I thought amazing I mean that was it really went in every direction that it needed to go I thought he did Mandy which has become one of my favorite films I think Mandy also goes it pursues the direction that it that it sets out in and and it perceives it to an extreme which I find to be much more interesting than something like dream scenario which I think has some interesting ideas the social death and explored in a supernatural way which as opposed to like a more typical drama exploring social death is interesting to me but I I didn’t think that it went far enough I think what could have been really interesting is if the film had played more with the dream aspect and moved directly into is it real or is it dream more and more of the film taking place within dream I think that could have been an interesting place to just explore but you know what it’s been 3 days and I’m still thinking about it so maybe I’m not giving it enough credit for what it accomplished anyway these are my 5 minutes of talking about Nicolas Cage films and talking to my computer and feeling very embarrassed about it the whole time thank you for reading or listening thank you computer for listening thank you reader for reading

Reflection

It appears that Windows voice to text when used out of the box with no settings tweaks eschews punctuation entirely, so that’s an impactful difference. This lack of punctuation really emphasizes my rambling. I do a lot of asides when speaking, more than I do when I’m writing, and especially more than I do when I’m doing more typical academic writing. Even in the process of writing this reflection I’m a very iterative writer, I re-read a lot, make tweaks, swap words. In oral communication that’s not an option. I’m itching to edit that text but am going to resist.

By and large the voice to text functionality captured the words I was saying. Without any punctuation it becomes difficult to parse, especially since speaking flows organically. It also doesn’t capture proper nouns to capitalize, which makes sense as a limitation of speech to text. I tested this in ChatGPT’s speech-to-text feature on their iOS app and it captured proper nouns to capitalize them. I think that is neat to reflect on even if it just the technology limitations of the default Windows speech-to-text versus a flagship consumer AI product.

I consider the lack of proper nouns a mistake as it makes reading the text more difficult. I’ve been writing a lot of alt text and image descriptions and using NVDA to navigate courses lately at work, and this has been something I’ve mulled over – proper nouns are contextually only important in written communication, but when they are missing I do find it disruptive. But I am able to follow when people tell me stories verbally without capitalization… interesting.

My vocal ticks like ums and hmms all seem to have been captured in different ways – as letters or other words but not typically captured as ums or ahs. I consider that a mistake as it’s changing the contents of the text. I wonder if it’s doing that not because it’s difficult to identify an um as an um, but trying to distance from the idea that the content is voice to text, as written text wouldn’t include ums and ahs. Trying to obfuscate that it’s a transcript of spoken word perhaps?

Had I scripted and rehearsed this I think there would be less asides, less wandering, and a much more concrete structure. Less errors from the vocal tisks being transcribed. Overall a much easier reading experience – but I would imagine less interesting to listen to! A space that would be interesting to explore is the contextual importance of some grammar rules like proper noun capitalization. It bothers me when reading a text that doesn’t have it, but it simply doesn’t exist for oral storytelling. On the flip side – what exists in oral story telling that doesn’t in written? I keep setting up and knocking down examples for myself – emphasis can be communicated through bold, italicizing, or size. Pacing: through…. punctuation. Characterization or accent can be dissected and written but I’ll admit I often have to read out loud to understand what is being said when accents are written in books (the Nac Mac Feegle spring to mind, to bring another mention of Sir Terry Pratchett to my writing in this course). I think that oral storytelling’s slippery, ephemeral nature is the starkest contrast to written storytelling. Written text is captured, oral storytelling resists that capture.

“We can plant ideas in each other’s minds using a finite set of words that we recombine into an infinite set of meanings” [00:43]

I’m thinking about this in the context of generative AI – a lot of the conversations around AI I’ve been involved in are situated around the idea of original production and what constitutes that. For example, can generative AI create new things or is it just remixing things – if all my experience with language is remix, what is remix, is remix creation? From an art copyright standpoint remix is creation after a point, this sentence centers the idea all language expression is remix.

“Whenever you utter a sentence, you are only uttering a tiny proportion of the information you know about the scenario” [07:15]

Thinking about text and technology, text as a construction, what you choose and select in terms of building materials and structure can result in many different spaces. Different places in the world with different contexts build homes in many ways – mitigating heat or trapping warmth, indoor outdoor living, or stark divides between interior and exterior. It makes sense that different places in the world would develop different structural needs for language in the same way.

“In the 1970s and 1980s in cognitive science it became essentially taboo, to think about how language might shape the way you think” [09:50]

This is fascinating to me because it’s one of those things that seems so obvious…  For example, when testing for reading comprehension you construct tests to explore a certain competency, but you can’t test what isn’t articulated and different languages would articulate different things – wouldn’t the difference result in differing priorities or pathways for learning?

Later Dr Borditsky speaks about “what we attend to” which I think is an interesting idea – sort of like the affordances of technology we need to consider what we are afforded by our use of language.

“The other thing that language does for us is helps us construe and construct events” [26:05]

Again, with these concepts of construction, assembly. This comes up for me a lot in work. I work a lot via text, emails, texts, MS Teams. It’s interesting to note in these written formats that I see a lot of this passive voice rearing its head in times when mistakes have been made. I do find myself deliberately constructing space between my team and events when I need to articulate there’s no causal link between a mistake and our team, for example. I’m also a guilty reader of online advice columns and when the passive voice creeps in you know the author has been up to some truly heinous misdeeds.

“But on the other hand, cultures also reduce cognitive entropy” [44:30]

This is connected to a Derek Bruff newsletter I read last week around creativity, divergent thinking, and AI. Bruff mentions that problem solving is generally a combination of divergent thinking (brainstorming) and convergent thinking (what of these options is the best) (Bruff, 2024). He indicates AI can be a wonderful tool for brainstorming but can also make it difficult to come up with out of the box ideas – like Borditsky mentioned once you have the tool or idea it’s difficult to come up one more level back to that divergent thinking, to reinvent the wheel so to speak.

“Depending on the level that you define universality, there are of course things that you find across all languages” [51:50]

Borditsky goes on to list some areas of universality but starts with the idea of human languages being learnable by humans – I find that a fascinating way of starting that response. This again takes me back to natural language processing and the increasingly blurry line between human and machine language. We’re in the midst of a really significant change to those boundaries and while I’ve been considering the impacts these changes will have on machine language, I think I’ve not been considering what might be coming back the other way in terms of impacts on human language. In the defining terms activity we thought about this – ways that words have been evolving in light of the internet.

References

Bruff, D. (May 16 2024). Creativity, divergent thinking, and AI. Intentional Teaching Newsletter.

SAR School for Advanced Research. (June 7 2017). Lera Boroditsky, how the languages we speak shape the way we think [Video]. YouTube. https://www.youtube.com/watch?v=iGuuHwbuQOg