Brennan Antone

| On the academic job market |
| Assistant Research Professor at Cornell, studying social networks in teams/organizations |
| he / him | 🏳️‍🌈

[Brennan Antone @brennan_antone]

With @OpenAI and competitors (HuggingChat, @poe_platform , etc.) making custom GPTs free to share and use, it will be interesting to see how makers find better articulated use cases for Gen-AI and build user experiences around them.

[Brennan Antone @brennan_antone]

It was great to present a tutorial/hack-a-thon on customizing Gen-AI at @icwsm with @whatsmalteupto last week. We examined creating specialized, sharable experiences (“apps”) out of general-purpose AImodels. Participants created wonderful AI experiences during the session!

[Brennan Antone @brennan_antone]

@decohere_ai Even if the human does not create images - Human judgement plays a key role in the process. Almost like the AI is a "brainstorming" tool for divergent thinking, while the human performs in the convergent part of an iterative design process.

[Brennan Antone @brennan_antone]

What is the right metaphor for describing the role of the human in Gen-AI? “Artist” seems wrong. “Curator” may be better.
@decohere_ai 's new Station feature emphasizes humans as selectors. It continuously generates alternative images, placing the user in a curation role.

[Brennan Antone @brennan_antone]

Our tutorial at #ICWSM will cover how to customize GPTs/LLMs to create specialized experiences online (AI "apps"), and how to examine AI experiences from a research perspective. with @whatsmalteupto

[ICWSM @icwsm]

🔥We are excited to announce the tutorials at #ICWSM 2024!
👉Check the list out at https://t.co/Sol8ObaSyT .

[Brennan Antone @brennan_antone]

Another Midjourney Image Jam event was held this week with @Grimezsz and Midjourney's @DavidSHolz

[DogeDesigner @cb_doge]

BREAKING: 𝕏 is in talks with Midjourney for a potential partnership.

[Brennan Antone @brennan_antone]

Similar results are possible with Python using pyodide in Quarto. Extensions for both created by @axiomsofxyz

[Brennan Antone @brennan_antone]

WebR is game-changing for teaching R!
With Quarto, you can render an HTML notebook that lets students write and run code in their web browser. No installation or servers are required. #webR #quartopub #markdown #rstats

[Saining Xie @sainingxie]

The key takeaway is from the "Emerging simulation capabilities" section. Before Sora, it was unclear if long form consistency could emerge on its own or if it required complex subject-driven generation pipelines or even physics simulators. OpenAI has shown that, though not perfect, these behaviors can be achieved with end-to-end training. Yet, two essential points have not been discussed.

1. Training Data: No talk about training source and construction at all, which might just imply data is likely the most critical factor for Sora's success.
👀Speculations: There's already much speculation about data from game engines. I also anticipate the inclusion of movies, documentaries, cinematic long takes, etc. Quality really matters. Super curious where Sora got this data from (surely not YouTube, right?).

2. (Auto-regressive) Long Video Generation: a significant breakthrough in Sora is the ability to generate very long videos. The difference between producing a 2-second video and a 1-minute video is monumental.
In Sora, this is probably achieved through joint frame prediction that allows auto-regressive sampling, yet a major challenge is how to address error accumulation and maintain quality/consistency through time. A very long (and bi-directional) context for conditioning? Or could scaling up simply lessen the issue? These technical details can be super important and hopefully will be demystified in the future (3/n)
https://t.co/NxZde24AHS

[Brennan Antone @brennan_antone]

@midjourney @DavidSHolz @OpenAI The first video is rendered with the Sora model for video generation, not footage pulled from actual Minecraft. The AI can simulate gameplay and control multiple characters.

Not perfect, but what does it mean to imagine virtual worlds or run simulation models with Gen. AI?

[Brennan Antone @brennan_antone]

@midjourney CEO @DavidSHolz always described the end goal of Generative AI models as not "image generators" per say, but as engines to simulate Virtual Worlds.

Now, @OpenAI #Sora realizes this idea, with the model simulating (FAKE) Minecraft.

Source: https://t.co/QPExaCywxg

[Historic Vids @historyinmemes]

Video made with the not-yet-released Sora Al technology just announced from OpenAi.

[Brennan Antone @brennan_antone]

Generative AI helps when programming – breaking down problems, suggesting packages, debugging, etc.

Here's a GPT (free for ChatGPT Plus) of how I adjust GPT-4 with instructions and custom knowledge when I work with R:
https://t.co/BhcKvsYpP4
#GPTs #r #Rprogramming #Rstats #GPT4

[Daniel Severo @_dsevero]

Best poster award

[🏀 Jimmy Xu 🏓 @MKTJimmyxu]

#webR will be a good substitute for the Rshiny server. Using webR, we can execute #Shiny applications entirely on the client side (e.g., the browser), harnessing the client's resources and presenting prospects for scalability. Resource for beginners: https://t.co/lyb69rTH5R

[Journal of Computer-Mediated Communication @ica_jcmc]

If you haven't already, please check out our recent special issue edited by Nancy Baym and Nicole Ellison on the Future of Work. Articles include:

[Brennan Antone @brennan_antone]

I've been doing ethnographic research on online communities about generative AI. So much community knowledge production contributes to our use of these otherwise black-box models. Human-Human interaction shapes Human-AI interaction. #midjourney #AIart #aiartcommunity

[Jasmine Wu @ywu2450]

✨Our new paper, with co-authors @brennan_antone, Leslie DeChurch, and @noshir, is out in the “Technology and the Future of Work” special issue of the Journal of Computer-Mediated Communication @ica_jcmc. It’s my pleasure to publish on my fav journal!

[Brennan Antone @brennan_antone]

Workshop materials are available online – See the handout ( https://t.co/6l1Gm7sRwk ) and the two slide decks (https://t.co/dEF6bvRAcK https://t.co/77Vu2uku77 ). Reach out if you would be interested in using any of the teaching exercises

[Brennan Antone @brennan_antone]

Excited to have presented a workshop at @SocNetAnalysts Sunbelt with @SonicNU . This workshop shared interactive teaching exercises for motivating social network thinking. Excellent presentations from @ktanaka086 @ywu2450 @vvsuschevskiy @noshir

[Brennan Antone @brennan_antone]

Lastly, we are launching Awesome Teaching About Networks, a curated list of teaching exercises and other resources for anyone interested in teaching network thinking: https://t.co/22fW3licAv
Feel free to share this list and suggest additional teaching resources we can promote!