AI wobbles. AI data stories. Rebuilding Gaza
Hello and welcome to “How To Speak Data”
“Wait a minute? Wasn’t this The Sweet Spot”??
Good question. It was?but the name didn't make sense to anyone!?So henceforth - It's "How To Speak Data"?
This month: a potpourri of fascinating things I’ve read recently.?I’ve also got a huge update of what I’ve been up to this summer, and what’s coming up.?
Our next Chart Chat is on Fri Sep 13. We're going to share the debates we've had so far while writing our new book. Expect fireworks. I have OPINIONs about text-only dashboards and population charts!?https://buytickets.at/datarevelations/1359750/r/acHero image for Chart Chat September 2024.?
Rocky times for AI?
Are we entering a difficult time for AI? The Economist explains how it is dropping towards the “trough of disillusionment” (referencing Gartner’s Hype Cycle nomenclature). It references the lost value of the leading AI firms, as well as mismatches in the capabilities that were promised and then delivered. All is, probably, not lost, says The Economist.?
On a somewhat different note, at Tech Policy, Maroussia Lévesque describes the “analog privilege… that people at the apex of social order” benefit from. Automations and AI gloss over nuances and contextual clues, leading to decisions that lock people into “algorithmic prisons”. It’s a sobering essay and one I saw described by prominent YouTuber ProZD. His most recent video explained exactly this issue: one of his videos was removed from YouTube search. A subscriber alerted him to the problem,? and he then spent weeks engaging with humans at YouTube to solve the problem. His video explains his privilege: he has nearly 4 million subscribers. That gave him the access to humans that almost everyone else would not have. Is this AI privilege something you have seen?
Can AI tell data stories?
The wonderful team at The Pudding asked “Can AI make a data-driven, visual story?” They use Claude to try and build a Pudding-style story, from ideas to implementation. Overall, they give Claude a C+. Claude is good, almost great, but certainly not as powerful as a data storytelling artisan.?
The Pudding team recognise that “good enough” might satisfy large audiences. We all might want our house’s furniture to be hand-crafted but most of us are satisfied with buying from IKEA. The Pudding conclude that AI’s data-storytelling skills are approaching the IKEA level.?
That's a little worrying, I think, for the artisan data storytellers. For now, It takes a domain expert to get functional, effective output from Claude, meaning we're still just about safe in our jobs. I think.?
I’d love to know your experiences. Let me know on LinkedIn how you’re using AI and whether you feel threatened by it.
Rebuilding Gaza
Reframing stories can bring new focus to an issue. Bloomberg has developed a powerful data story about rebuilding Gaza. It’s one of the most moving and effective data stories I’ve seen this year. Talking about the urban destruction helps me see this terrible situation in a new way.?
If one critique was to be allowed one critique, it would be this. The 3d elevated bar map (see above) shows how much rubble there is across the Gaza strip. It's powerful, in no small part because the bars are so high. But those heights are are arbitrary. What would happen if the highest bar was a third of the height? A tenth? Is the extent of rubble being misconstrued because of the height of the highest bars? I’d have loved to see a measure of the percentage of buildings destroyed to get a better sense of the completeness of the destruction. Despite that small quibble, it's a great piece of data journalism.
What's happening?
I'm also speaking at multiple Tableau User Groups soon:
Register for these or others at the Tableau User Group page.?
That's all for now! Any comments, let me know!
I like the new name!
???????? ???????? ???????????? - Analytics CoE | Playtika
2 个月Andy Cotgreave I'm a big fan of your work and even got excited to see you live at the last TC BUT... I can't admire the visualization you are referring to as an "effective data story" as it is missing a huge part of the story. the story of the atrocities happened on Oct 7 in the south of Israel. As I don't want to go too much into politics, I will try and stay in the professional data viz area and share a few examples created by Israeli creators: "Oct-7th Bloody Saturday" by ?Nir Smilga?? https://public.tableau.com/views/Oct-7thBloddySaturday/7thofOctoberattacksmap?:language=en-US&:sid=&:redirect=auth&:display_count=n&:origin=viz_share_link "Bring Them Home Now" by ?Nir Smilga?? https://public.tableau.com/app/profile/nir.smilga/viz/BringThemHomeNow/BringThemHome "Foreign Casualties on Oct-7th Terrorist Attack" by ?Nir Smilga?? https://public.tableau.com/app/profile/nir.smilga/viz/ForeignCasualtiesonOct-7thTerroristAttack/ForeignCaualtiesonOct-7thTerroistAttack and finally, as we are talking about the media coverage of the situation: https://public.tableau.com/app/profile/nir.smilga/viz/Israel-PalestineMediaBias/IsraelGazaMediaBias I would hope you will review these viz's to get a "fuller" understanding of the issue.
Co-Founder & CEO at Data Literacy
2 个月I like the new name! ??
Ex-UN Data Specialist | Tableau Ambassador | ESG Expert #VisualClimate | Founder @Visual Climate
2 个月Andy this is marvelous and well structured, very informative. amazing