Tom Cramer

I don't always tweet goodbye to dear friends and colleagues, but when I do it's to @jessiekeck

Is there any better way to spend a Friday morning than chatting about DAMS and Cloud with @derivadow @alexwlchan @_tuan_p_ @CaylinSSmith @richard_d_jones @cgknowles @anjacks0n @NeilSJefferies @clemo2021 ? It is a treat to be part of, and benefit from, the great hive mind

@justin_littman Can have it still have APIs? :)

After a few decades at this, here's a conclusion: If you can use digital library software out of the box (just config, no coding) then great. Otherwise, roll your own that does only / exactly what you need w/ Django, Rails or similar. It will be more maintainable, less code.

Let us reflect on how complete and generous the Smithsonian’s release of millions of images is: 1) CC0; 2) easy-to-use API; 3) full metadata available in GitHub; 4) IIIF viewer + portability; 5) images are downloadable at very high resolution. This is a great model—kudos to SI.

A (new) definitions blogpost Responsibility, Accountability, Transparency (trust, brittleness, explanation) joanna-bryson.blogspot.com/2019/12/respon…

.@teemu_rooos Elements of AI introductory course (230,000+ registered participants) has a goal to "help people feel like AI is their thing" -- not something foreign

.@teemu_roos Self driving cars get the hype, but AI-based social media filters and recommender systems are shaping our world much, much more (and already in place) #fantasticfutures2019

.@teemu_roos leads off the #fantasticfutures2019 workshop day with the fundamentals of AI--the most basic concepts "because we need a common foundation of understanding AI" first and foremost. These seems especially true in LAMs

@khetiwe24 @gworthey @Aslaks More / better MARC is surely just a stepping stone to getting weirder. (I am thinking of the Ford quote about faster horses here...)

@gworthey @Aslaks Definitely. Challenges, though: for ex, when cataloguing has not been done at item level, but digitization is. Do we prioritize making MARC with AI or can we be weirder? (Both, obvs, but trying to imagine many solutions here...) #FantasticFutures2019

Nice comments from @Aslaks on not being a slave to the catalog when envisioning opportunities for “weird intelligence” and AI in #glam #fantasticfutures2019

@1n9r1d @kcimc @tvdmaps Yes, this. As a profession we tend to stick with text and lists. A concerted effort to go beyond ourselves is in order...

@tcramer @kcimc @tvdmaps How about getting outside of our own heads, practices, and collections; and instead using the opportunity to experiment with AI and particularly ML to reconnect with our communities? #FantasticFutures2019

The talks from @kcimc and @tvdmaps at #FantasticFutures2019 have cemented my growing suspicion that the future of AI will only be realized via new interfaces; without different ways to generate, view and think about AI, libraries will focus on redoing what we already do

@kcimc @tvdmaps FYI, lots of novel UIs (including those shown in his talk) on @kcimc's page at kylemcdonald.net

They never discuss digital at the National Library of Norway. It’s just assumed as an integral part of the work. #FantasticFutures2019

@khetiwe24 @c_a_demarco What kind of metadata would you add back into MARC? Trying to extrapolate from building/not-building or rail line vector data examples and struggling

@c_a_demarco I agree! We're planning to incorporate this in MARC records so there is as little barrier to discovery in the catalog as possible.

This talk makes me want to reenroll in college and take a class from @j2bryson #fantasticfutures2019 Amazing job of framing issues large and small, and rigorous definition of AI and social issues.

? sparked by #fantasticfutures2019 training talks: what's the best way to organize a library for AI projects/work? A little of a lot of people? A skunkworks AI lab? Other? @math_rachel likens this to corporate AI question--embed data scientists in teams or have stand alone unit

What is AI? It’s this per ⁦@j2bryson⁩ #FantasticFutures2019

I wonder what it would like to get 1% (or 10%) of the world's librarians to be AI-savvy. After @math_rachel and @teemu_roos inspiring talks, it doesn't seem like a crazy goal. #fantasticfutures2019 #ai4lam

AI myths are dangerous, per @teemu_roos. If you avoid AI, significant opportunity costs in progress. But more importantly, ignorance prevents discourse and addressing social/policy issues head on #fantasticfutures2019 #ai4lam

.@math_rachel Thank you! A slow-to-post follow up question: you mentioned FastAI part-time students several times--do you see a role for part-time AI practitioners (e.g., a librarian), or is it practically something best done as full time work? #fantasticfutures2019 #ai4lam