How did you experience catching stalls alongside GAIA?
I would like to hear about how your experienced Stall Catchers with a bot active on the platform and what you think about GAIA🤖 in general!
Your insights will contribute to Human Computation Institute’s understanding of human/AI partnerships and will inform my PhD research.
In my cultural anthropological PhD research I investigate human computation-based citizen science projects and the meanings these projects have for the involved parties (like participants, designers and developers). I’m particularly interested in forms of human/human and human/machine collaboration, the different role allocations within these projects and how they affect our everyday lives.
If you would like learn more about my research and the perspective I take you are welcome to contact me via email: l.veprek@lmu.de
You can share your experience here in the forum or send me an email if you would like to tell me about it personally.
Please note that this forum is public. If you prefer to contribute in an anonymous way, please contact me via email. Any information you share with me personally will be anonymized and used exclusively for research-related purposes.
Hi, Libuše. I already shared some initial impressions of GAIA with you, and then later with Pietro in the ‘Final Hour Hangout.’ It was nice that GAIA was very deliberate (relatively slow). While her skill level was generally pretty good, she was not on par with our best catchers. That combination of speed and skill, I felt, made her a relatively non-threatening first encounter for those of us unfamiliar with bots as collaborators. I’m looking forward to seeing the next iteration, and how we can leverage AI to improve stall-catcher productivity.
Mike C.
Hi Christiane! Thank you very much for sharing your viewpoint! That’s great to hear and I’m sure there will be further opportunities for GAIA!
Best, Libuše
Gaia was the my biggest concern during the catchathon. The fact, she was capable of undertaking the task without having to take any rest, was the biggest advantage she had over us. That was enough pressure and encouragement to keep me going. If not for the initial hitches that prevented her from performing continuously during the event, I’m certain she would have beaten me. Did notice that her sensitivity level wasn’t 100%. She should be at her best for the next event. It was fun and challenging, competing against her.
Now that the new bot study with three bots is underway, I would like to invite you to share your impressions here in the forum.
Your insights will contribute to Human Computation Institute’s understanding of human/AI partnerships and will inform my PhD research.
If you would like learn more about my research and the perspective I take you are welcome to have a look at this blog post or to contact me via email: l.veprek@lmu.de
Please note that this forum is public. If you prefer to contribute in an anonymous way, please contact me via email. Any information you share with me personally will be anonymized.
Here’s the latest information from Pietro Michelucci, director of the HCI, on the bot study (copied from the Stall Catchers chat)
"Hey Catchers! I just wanted to explain something about the bots and the leaderboard.
If you give a bot a more powerful server, it will climb the leaderboard faster. And we gave these three bots, very powerful servers to use.
The reason we did that is because for the research question we are trying to answer, we would like the total number of vessels annotated by the three bots to be about equal to the total number fo vessels annotated by all humans combined.
Since there are many humans playing, and only three bots, we need to the bots to go faster for this study.
It’s actually pretty amazing that we’ve had some humans - supercatchers - climb past these superbots on the leaderboard over the last few days.
The powerful servers being used by the bots are not cheap - and for this reason, any future bots that become regular catchers, would likely not be so insanely fast."
Hi, Libuše. Well, GAIA feels rather “familiar” since we’ve previous experience with her. She seems to be at the same average skill level and playing speed as before. I have initial impressions of the other two bots’ skill and speed, but really - after a while - don’t think too much about them. Perhaps because we don’t know their ‘pronouns,’ and certainly we can’t greet or interact with them in the chat box. One of the newer bot’s playing speed seemed to change (slow) significantly at one point, which made me wonder if their handler was doing some tweaking, and whether or not they bothered to “redeem” points.
It is easy to outdistance all three bots at the beginning of a new SC day, but with the knowledge that they will likely be and stay ahead of you when you awake in the morning - unless you care to give up your regular life and/or some sleep. Still, I enjoy seeing them doing their thing at or near the top of the leaderboard, knowing they are helping to advance the science and our development of bot-human collaboration. Maybe if their handlers ever care to use the chat box or forum, the bots can acquire some additional personality by association - not that they need ‘personality.’
Or, we could ‘meet’ their creators in the blog, perhaps. I think we did already meet GAIA’s creator that way, but don’t recall if we did the other two. If so, you could send the link for those of us who are new or (like me) might care to refresh our memories. Best to all, and go team “Crowd Bots.”
Mike C.
Thank you so much for your valuable feedback, @caprarom!
That’s a very good point. As far as I know the other bots and their creators haven’t been properly introduced so far. So I will forward your request
Currently, you can only learn a little bit about the differences between the bots in this blog post: Bots, that are going to play Stall Catchers along humans
I enjoy playing with the bots. I even have a favorite bot (Gaia). I think it is because she is the first and she has a bit of personality. I reserve my final opinion about the bots pending the results of this study. Do they meet the team’s objectives for creating these bots? If yes, then they will be a welcome addition for analysis of large data sets. After all the goal of this citizen science project is to help out a lab focused on Alzheimer’s research. If they help the lab achieve their goals, then great.
Adding some “personality” to the bots might also be helpful considering that some of the teams on StallCatchers are composed of school science classes. Middle school students might enjoy playing with a bot that has some personality. The bots chats could even be a teaching tool to educate kids about Alzheimer’s research, neuroscience, and AI.
One of the newer bot’s playing speed seemed to change (slow) significantly at one point, which made me wonder if their handler was doing some tweaking, and whether or not they bothered to “redeem” points.