Monthly Archives: April 2023

[Explored] Mairie de Londres – City Hall of London flickr photo by Jopa Elleul shared under a Creative Commons (BY-NC) license

One interesting thread (via Kate T) that emerged in a discussion in ETMOOC2’s inquiry into Artificial Intelligence and the emergence of ChatGPT, Bard and others was the impact on the environment. I hadn’t really thought much about it (I know, I should have) but when we think of how cryptocurrencies and NFTs all came to have a rather outsized impact via energy use and computing power, it made sense that generative AI would likely be doing the same.

An article in my May 2023 Wired magazine entitled “Generative AI’s Dirty Secret” by Chris Stokel-Walker also explores the topic, and while data from companies like Microsoft and Google and others releasing AI into the world is hard for outside researchers to come by, the article notes that “AI-fortified search engines are likely to demand far more computing power, which means a massive increase in energy use and carbon emissions.” (Stokel-Walker).

I suspect these companies are already working on data center efficiencies and other alternatives to the way the algorithms work to combat this — at the least, conserving energy will save them money, but also, many of these companies have environmental, climate-focused plans in their mission statements, although how much we should trust those mission statements in a business model designed for profit over the common good remains to be seen.

Kate shared these three articles that might be of interest on this topic:

• AI’s Carbon Footprint (via MIT Technical Review): https://www.technologyreview.com/2022/11/14/1063192/were-getting-a-better-idea-of-ais-true-carbon-footprint/
• Energy Consumption (via Mark Labbee, at Tech Target): https://www.techtarget.com/searchenterpriseai/feature/Energy-consumption-of-AI-poses-environmental-problems
• Risks and Benefits (Environmental Science & Technology): https://pubs.acs.org/doi/10.1021/acs.est.3c01106

This diagram — which widens the scope a bit — is from the article at Environmental Science & Technology:

As with all things, technology-related, we have to keep an eye on the picture of the world at large, and consider the impacts on the environment and climate, and us, as we dive into new tools and consider the pros and cons.

Peace (And This Planet),
Kevin

(This is the fourth inquiry experiment this week with Artificial Intelligence for ETMOOC2. The first was a poetry exchange between AI platforms. The second was an Interactive Fiction project with ChatGPT and Twine. The third was generating code for a website about saxophone design).

This was a strange idea that came to me during the week, but I wondered how ChatGPT would do if I asked it to make a Mad Lib-style story (with prompts for words that a reader/player would suggest and fill in before reading the actual story) on the theme of ChatGPT. This concept became a starter point for my inquiry activity.

I actually ended up asking it compose me four different stories — one about The Day In The Life Of ChatGPT. Another is about a conversation between Google’s Bard and ChatGPT (which it wrote, without my prompting, emerged in the style of Shakespearean English — sort of). ChatGPT came up with that title all on its own: A Hilarious Conversation Between Google’s Bard AI And ChatGPT. (Note: the “hilarious” is an oversell, in my opinion). The other two stories involve a day without technology called When Nothing Worked and finally, Walking A Robotic Dog.

You can play the four Mad Libs.

I then reversed course and had ChatGPT play one of its own Mad Libs, asking it via a series of queries for random Parts of Speech words for each of the prompts it had just generated for the story (Such as: give me a random “noun”), plugging those words into the Mad Lib story, and the final result came out predictably strange (but perhaps no stranger than any regular human Mad Lib game).

Process Notes: How I did this experiment was that I took the story generated by ChatGPT with a simple prompt — Make me a Mad Lib style story about … —  along with the list of word prompts that were generated by ChatGPT from my query and I moved those over into a Mad Lib generator hosted over at Flippity (which I use with my students). The formatting on the Bard/Chat story dialogue got funky on me and I can’t figure out how to fix it, so it became an odd run-on conversation. I added some character line breaks to help but it still looks mangled to me.

By the way, when I took the same prompts for Mad Lib stories into Google’s Bard, the results were awful and not at all useful. There really is a noticeable difference between the two AI platforms at this point for the kinds of inquiry that I am doing here, where ChatGPT has much more flexibility and creativity built in (or maybe it is more allowed to flourish, likely) than Bard has at this point in time.

Reflections: When we do Parts of Speech in class, we often play and create Mad Libs as a way to play games with language. I suppose ChatGPT could be a collaborator in the story element of these activities, but its sense of humor is only iffy, and what Mad Libs really need is a subversive streak to them. No need to even invite Bard to the party, though. One thing I noticed: The AI was limited in the Parts of Speech it would leave out of the stories.  It was all nouns, adjectives and verbs, and that got pretty boring as a reader/player. A human collaborator would have to expand the choices.

Peace (Filling In The Blanks),
Kevin

(This is my third experiment this week with Artificial Intelligence tools for ETMOOC2 – the first experiment was about bouncing poems off different AI platforms and the second was using ChatGPT to build out an Interactive Fiction game of choices).

In my further deeper dives with ChatGPT, I wondered about its ability to spit out code for a simple website. I know other folks are doing it, and using it to check code, but I was curious about what it would create for me, in terms of coding out a simple website.

I asked it: Can you give me the basic code for a website that is dedicated to the art of designing and making saxophones?

In seconds, my screen was filling with HTML code, with parting advice to me:

“This code includes a header with a navigation menu, a main section with information about the website and its features, and a footer with copyright information. You can customize this code to fit your specific needs and design preferences.”

Got it!

But what did the website look like beyond the raw code?

I asked ChatGPT how I could test its code and see what the site would actually look like to a viewer. The AI quickly provided me with some helpful advice on using a Text Edit application on my computer, saving the file as HTML and then opening the file up with a web browser.

You can take the HTML code provided and use it to create a website by following these steps:

1. Choose a Text Editor: You’ll need a text editor to write and edit HTML code. You can use a simple text editor like Notepad (on Windows) or TextEdit (on Mac), or you can use a more advanced code editor such as Sublime Text, Visual Studio Code, or Atom.
2. Create a New HTML File: Open your text editor and create a new file. Copy and paste the HTML code into the new file.
3. Save the HTML File: Save the file with a “.html” extension, for example, “poetry_celebration.html”. Choose a location on your computer where you want to save the file.
4. Open the HTML File in a Web Browser: Open your web browser (such as Google Chrome, Mozilla Firefox, or Safari) and go to “File” > “Open File” (or “Open” in some browsers) and select the HTML file you just saved. This will open the HTML file in your web browser, allowing you to see how it looks and test its functionality.
5. Customize the Website: You can customize the website by modifying the HTML code to suit your preferences. For example, you can change the text, add images, adjust the layout, and customize the styles using CSS.
6. Publish Your Website: Once you are satisfied with your customization, you can publish your website to make it live on the internet. You can do this by purchasing a domain name and web hosting service, and then uploading your HTML file and any other associated files (such as images or CSS files) to the web server provided by your web hosting service. This will make your website accessible to others via its domain name.

Note: If you are new to web development, you may find it helpful to learn more about HTML, CSS, and web hosting to better understand how to create and publish a website. There are numerous online resources, tutorials, and documentation available to help you learn these skills.

For some reason, Chrome and my other browsers didn’t want to play along (maybe I saved the file wrong?) and all I got when I opened the file was the code itself. So I found an online site called Liveweave to check it out.

Pretty basic, as expected, but certainly it could be a starting place for someone working on a functional, if unexceptional, website. I also wonder about whether asking it do something in Python would allow for more complexity and interactivity, which is really what I am after. That’s for another day …

Peace (and the Web),
Kevin