Profile

CommunityPlatform_1350x900.jpg

Brian Basgen

Edit My Profile

My Content

1 to 20 of 42 total

RE: Stanford University: The AI Index Report: Measuring Trends in AI

Posted By Brian Basgen 04-17-2024 03:43:35 PM
Found In Egroup: AI \ view thread
Thank you for sharing this Heather. Every so often an AI resource comes along that I think is worth stopping for and fully absorbing. I plan to spend significant time digesting this report. I think spending a few weeks on this will get me months ahead of the minutia. The report is thoughtfully constructed ...

Consortia approach to LLM?

Posted By Brian Basgen 04-01-2024 09:54:00 AM
Found In Egroup: AI \ view thread
Hi Everyone, Is anyone aware of an effort towards a higher ed consortia agreement with an LLM provider? OpenAI is offering enterprise licensing for large organizations, but this doesn't fit our need. I've also heard that Amazon is trying to make private cloud LLMs, but this also seems like a specific ...

RE: Administrative Systems Management Community

Posted By Brian Basgen 02-25-2024 05:37:46 PM
Found In Egroup: Administrative Systems Management \ view thread
Thanks for trying to re-invigorate the community, I hope it works out well! I'm probably not the intended audience for this group, so take this with a grain of salt. One example that could be an inspiration is a mailing list out of Haverford for "Workday in Small Schools." In this group, I imagine ...

RE: Sora video generation

Posted By Brian Basgen 02-17-2024 09:17:09 AM
Found In Egroup: AI \ view thread
Good point! There are several other interesting video products like Runway's Gen2, Morph Studio, Stable Video. To date, these products have made cool iterative advances that have pushed the field forward. The struggle to date has been that these tools tend to present a fairly flat world with limited ...

Sora video generation

Posted By Brian Basgen 02-16-2024 12:55:00 PM
Found In Egroup: AI \ view thread
Hi Everyone, If you haven't yet seen it, OpenAI is about to change the world again. https://openai.com/sora The technical paper is very accessible and enlightening. (https://openai.com/research/video-generation-models-as-world-simulators) -------------- Brian Basgen (he, him, his) Chief Information ...

RE: Using AI for image generation without copyright concens

Posted By Brian Basgen 02-01-2024 01:10:26 PM
Found In Egroup: AI \ view thread
Hi Lynn, Adobe Firefly is the only image gen AI I've heard of that has been copyright conscious, and, it is something Adobe is uniquely situated to deliver on because of their massive library of stock images. Thus, I think it is safe to say that, for the time being Adobe, is the only product on the ...

RE: LLM rankings

Posted By Brian Basgen 12-18-2023 08:57:13 AM
Found In Egroup: AI \ view thread
One of the three benchmarks (MT-Bench) uses GPT4 for grading. In general, one of the nice things about LLMs is that they can be good at evaluating its own content; e.g. when prompted after making a mistake they can often identify it since generating content is distinct from analysing content. The paper ...

LLM rankings

Posted By Brian Basgen 12-15-2023 01:54:00 PM
Found In Egroup: AI \ view thread
Hi Everyone, If you have wondered about the best LLM to use or keep your eyes on, this leaderboard sponsored by Hugging Face is one way to look at the landscape: https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboard It usefully contrasts GPT4 vs Claude, and it is interesting to see the viewpoint ...

RE: WCTE: Developing Institutional Level AI Policies and Practices: A Framework

Posted By Brian Basgen 12-11-2023 05:41:06 PM
Found In Egroup: AI \ view thread
Thanks for sharing this Heather. This is the best one I recall seeing: it is comprehensive yet concise! -------------- Brian Basgen (he, him, his) Chief Information Officer20 Park Plaza Building Emerson College | 120 Boylston Street | Boston, MA 02116For IT assistance, please visit: IT Help Desk

RE: What are the implications of truly multimodal models for education?

Posted By Brian Basgen 12-11-2023 09:14:42 AM
Found In Egroup: AI \ view thread
One of the ways I've used the multi-modal capabilities of GPT-4 is in working through historical images for a book I've been writing. Working with GPT, I've been able to identify things like the name of a naval ship based off just an image and scant details, details about tools being used in an early ...

Great gem in the online conference if you missed it

Posted By Brian Basgen 11-01-2023 04:07:17 PM
Found In Egroup: CIO \ view thread
I was happy to see the in person conference is now coupled with the online conference, but I didn't realize that the online conference materials can be accessed until November 16th! It has been hard to make the time, so this has been great. In case you may have missed it, the General Session for the ...

Community rule proposal about AI use

Posted By Brian Basgen 10-27-2023 01:38:00 PM
Found In Egroup: CIO \ view thread
I would like to propose a new rule for this mailing list, in the spirit of rules like non-solicitation, in light of a recent message on this list that disseminated problematic AI output: Any information that comes directly from AI should, at minimum, state what AI it came from. Bonus points if it includes ...

RE: AI Policy

Posted By Brian Basgen 10-24-2023 01:28:51 PM
Found In Egroup: CIO \ view thread
Hi Dan, I'm curious if the example you shared was written by AI? The formatting has inconsistency and the content varies between statements of policy to values to action items. It also demonstrates a level of brevity that is common with AI output. -------------- Brian Basgen (he, him, his) Chief Information ...

RE: Response Techniques for Online Harassment of Faculty/Staff

Posted By Brian Basgen 10-23-2023 05:23:37 PM
Found In Egroup: CIO \ view thread
Hi Steve, I suggest you take the question to the REN-ISAC community where folks can talk a bit more openly about their experiences and effective counters. -------------- Brian Basgen (he, him, his) Chief Information Officer 20 Park Plaza Building Emerson College

RE: Introduction and request for guidance

Posted By Brian Basgen 10-16-2023 05:10:18 PM
Found In Egroup: AI \ view thread
It might be fair to say that the number one problem with AI right now is a lack of privacy controls. When considering the importance of training data, gathering data sources seems to be a top priority of these tech companies. I've heard a few prognosticators suggest that enterprise AI (e.g. without privacy ...

RE: 2023 EDUCAUSE Horizon Action Plan: Generative AI

Posted By Brian Basgen 09-30-2023 05:56:32 AM
Found In Egroup: AI \ view thread
Hi Everyone, I read the AI summary in this thread and then the Horizon report. Be advised that the AI summary is misleading: it made it hard for me to understand the report due to its inaccuracy. I've noticed this with a few other articles and I think it would be helpful to have a discussion about ...

RE: "Instead of policing students, we need to abolish cheating" - Chronicle article

Posted By Brian Basgen 09-15-2023 01:59:57 PM
Found In Egroup: AI \ view thread
Thanks for sharing Danny, and your summary honed in on helpful points. The article is helping me realize that, as you say, this is going to be a long process with faculty, and that well reasoned perspectives like this one will help. At Emerson, we are focused on storytelling and creativity, and I suspect ...

RE: TII AI-detection

Posted By Brian Basgen 09-15-2023 09:11:18 AM
Found In Egroup: AI \ view thread
Hi Emory, I recommend reading Open AI's recent summation of the research on this subject, that detectors don't work: https://help.openai.com/en/articles/8313351-how-can-educators-respond-to-students-presenting-ai-generated-content-as-their-own -------------- Brian Basgen (he, him, his) Chief Information ...

RE: Math and the Physical Sciences - GenAI Use Cases and Insights

Posted By Brian Basgen 09-12-2023 07:13:00 AM
Found In Egroup: AI \ view thread
Sorry Heather, I should have provided more information. If you go to the domain https://www.messengerx.io/, the landing page has a few telltale signs. The site features many AI chatbots for AI girlfriends and boyfriends. Other chatbots appear to be solicitations. They have a few "normal" looking chatbots ...

RE: Math and the Physical Sciences - GenAI Use Cases and Insights

Posted By Brian Basgen 09-11-2023 01:12:51 PM
Found In Egroup: AI \ view thread
Thanks for the corrected link. messengerx.io raises an important question about what kinds of AI should be used/recommended. This domain gives the strong appearance of a scam/malicious domain, and I would not trust using it in any capacity. It is interesting because scammers don't always operate in ...