iiinova — where your digital self keeps evolving

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 18/03/2025

I kinda like ChatGPT’s approach to the question “who was I in another life?”. Mapping uncharted territories sounds like fun! Uncovering hidden patterns and structures, even better! #chatgpt #llm #prompt

010

Reposted by Nicolas F. Chaves-de-Plaza

Historici vertellen @historicivertellen.bsky.social · 10/03/2025

Goedemorgen! Mijn naam is Ramona Negrón (@ramonanegron.bsky.social), postdoc bij het KITLV in Leiden. Ik houd me vooral bezig met vroegmoderne koloniale slavernijgeschiedenis. Deze week deel ik hier wat mij zoal bezighoudt. Waar zouden jullie graag meer over willen weten?

149611

Reposted by Nicolas F. Chaves-de-Plaza

whet moser @whetmoser.com · 08/12/2024

i feel like one of the disappointments of current tech is that it frees you from mental tasks only to realize you weren’t going to do anything better with that mental processing

101559

Reposted by Nicolas F. Chaves-de-Plaza

Brandon Amos @bdamos.bsky.social · 05/12/2024

hi everyone!! let's try this optimal transport again 🙃

232931

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 05/12/2024

One of the reasons I opted for a PhD instead of a more traditional path in tech is to have access to, what I perceived were, more complex problems

010

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 04/12/2024

Could be. If they apply a transform to derive a new dataset. What is data viz for you?

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 04/12/2024

I think the action data/information can take is to steer your own mental model. Similar as to when two humans interact. The effect of the interaction might not be visible but it’s there (in their brains).

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 04/12/2024

Sometimes decision will be made based on this information. Hence the need for clear communication, which acts as a steering mechanism

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 04/12/2024

Note in this cases mental models are not limited to humans. A database would also be a mental model (albeit an externalized one)

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 04/12/2024

It’s all about designing interfaces to understand what’s going on in a mental model and, if needed, updating it

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 04/12/2024

The term human-data/information interaction captures more clearly what visualization means to me #dataviz

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 02/12/2024

The main implication of this observation is that current visualization systems might not be designed for how we more often reason … perhaps we need more multi-modal interactions?

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 02/12/2024

For instance, the “Which alloy combination results in the strongest material?” research question can be easily put in words. Can it also be easily drawn?

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 02/12/2024

1/n) Written language seems more suitable to express hypotheses than visual means #dataviz

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 29/11/2024

Nice chart! The colors threw me off a bit. Would have expected yellow is farther away (as in my mind you go from red -> yellow -> green)

110

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 28/11/2024

A big shoutout to my co-authors. Your support made this milestone possible! I learned so much during this two-year journey—especially about conducting user studies, workflow analysis, and clinical simulations. These are skills I look forward to building on in the future.

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 28/11/2024

Key findings: 💡 DEDS can save time if per-slice analysis remains efficient. 💡 Clinician workflows & user-centric design are crucial for adoption. We hope this study encourages more user-focused AI development in healthcare.

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 28/11/2024

Conducting this research involved: - A user study with clinicians at Holland Proton Therapy Center - Developing a simulation model to evaluate workflows - Assessing time savings while balancing delineation quality

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 28/11/2024

This research combines AI & human-centered approaches to tackle challenges in radiotherapy. We studied how delineation error detection systems (DEDS) can help optimize workflows for clinicians dealing with AI-generated tumor and organ delineations.

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 27/11/2024

The production level of @deep-mind.bsky.social videos is crazy. They don’t fail to motivate me in my own scientific pursuits. www.youtube.com/watch?v=mGsy...

youtube.com

A New Age of Opportunity | AI for Science Forum

YouTube video by Google DeepMind

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 22/11/2024

Should we try to (and could we actually) follow this design principle when developing technology? (E.g, self replicating machines)

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 22/11/2024

In contrast, evolution has produced fault tolerant machines that contain within themselves anything needed for their replication.

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 22/11/2024

Technology and knowledge is incredibly fragile. Delete a couple of hard drives and most of what we know how to do is gone. Right?

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 22/11/2024

Already looking forward to seeing the submissions next year. Super cool challenge (and slick logo 👌🏻)

010

Reposted by Nicolas F. Chaves-de-Plaza

Tim Gerrits @dertimgerrits.bsky.social · 22/11/2024

The 2025 SciVis Contest Data is available! If you/your team/your students are looking for an exciting visualization challenge (it's about helping scientists develop new, sustainable materials), take a look at the contest data: sciviscontest2025.github.io #Visualization #IEEEVIS #SciVis #Contest

143

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 22/11/2024

From experience: #AI and #DataViz researchers need advanced domain understanding before building effective tools. Why not empower domain experts to create their own tools and tackle their research questions directly? Programming languages are an example of this idea, but can be too complex

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 22/11/2024

Listening to Demis Hassabis from @deep-mind.bsky.social talk about the future of applied AI research, specially for biology, fills me with energy and hope for the future www.youtube.com/watch?v=nQKm...

youtube.com

AI for Science with Sir Paul Nurse, Demis Hassabis, Jennifer Doudna, and John Jumper

YouTube video by Google DeepMind

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 22/11/2024

Super nice prints! Gives me a nineteen century botanical illustration vibe

010

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 20/11/2024

Oh very cool. Are feeds like lists in Twitter?

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 19/11/2024

Epistemology and entanglements: two words that stayed with me from today’s very refreshing talk by @gotdairyya.bsky.social at TU Eindhoven. It’s always good to take a conceptual step back to find opportunities for improving and strengthening #dataviz practice 💪🏻 go check Derya’s papers to learn more!

020

Reposted by Nicolas F. Chaves-de-Plaza

Andy Kirk @visualisingdata.com · 14/12/2023

What Giorgia Lupi has and is going through with Long Covid is horrendous, I am so sorry for her facing all this. In this piece for New York Timed she reflects on her experience with her typically exquisite flair www.nytimes.com/interactive/...

nytimes.com

Opinion | 1,374 Days: My Life With Long Covid

Chronic illness has a way of picking apart your mind and breaking your heart.

31518

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 08/12/2023

We are witnessing a dramatic increase in the size of LLMs like ChatGPT. No one knows how big is big enough. Shouldn’t we instead try to attain human-level performance while keeping the model as concise as a human brain? How big is the human brain, actually (in LLM terms)?

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 24/11/2023

I couldn’t believe when they told me ChatGPT could reconstruct what I -mis-typed without looking at the keyboard … but it can 🤯, ChatGPT is also the best keyboard auto-correct it seems …

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 16/11/2023

If a task cannot be successfully fully automated, the user will need to play an often boring and fatiguing supervisory role. In these cases, a human-centered interactive intelligence approach should be favored …

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 16/11/2023

Like the term “interactive intelligence”

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 16/11/2023

A limitation of ChatGPT is that it cannot leverage models. For instance, it might “know” of e=mc^2. But it cannot compute with it. Therefore, there is a lot of knowledge currently being wasted…

010

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 30/10/2023

Is anyone aware of work that performs a user evaluation of the scalability and performance of spaghetti plots for ensemble analysis tasks? Intuitively, with many lines they become too cluttered and their effectiveness reduces. But haven’t found a reference for this …

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 28/10/2023

Also, I have no idea where the vis community is nowadays. Is it X, LinkedIn, Mastodon? I feel it is sooo fragmented 🥲

100

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 28/10/2023

Nice collection and perspective on some of this year’s #ieeevis papers

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 26/10/2023

Nature is like a genie that answers exactly the question we pose, not necessarily the one we intend to ask. But we have to believe […] that the answer to the question we wish to ask does exist in nature. - Judea Pearl

000

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 25/10/2023

One goal of causal inference is to create a smoother human-machine interface, which might allow the investigators' intuition to join the belief propagation dance - Judea Pearl

030

Nicolas F. Chaves-de-Plaza @nicodeplaza.bsky.social · 25/10/2023

Instead of data visualization why not think more in terms of human-data interaction? This makes it easier to think of text-based UIs like ChatGPT’s as part of the communication toolkit

020