Tools, technologies and techniques: share and tell
Tool Gauge, Chris Makrenos (c. 1939)
Our next Hacks/Hackers Brisbane meetup is an opportunity for everyone to come together and share any interesting software tools and technologies they have used or discovered recently. Or to share any long-term trusted technologies that may be lesser known amongst the community.
Whether you're a journalist, a developer, a designer, a data scientist, or anyone who is interested in the intersection of technology and journalism, this meetup is for you. We don't need you to prepare anything significant in advance – just be willing to give a quick 3-minute intro to the tool and how you are using it.
This is a great chance to learn from your peers and discover new tools and techniques that you can apply to your own work. We'll have a projector set up that you can use if required, and we will leave plenty of time for networking and discussion, so come ready to share your insights and connect with like-minded individuals.
So mark your calendar for Wednesday 3rd May, and join us for an evening of learning, sharing, and fun! We look forward to seeing you there.
When: 5:30pm, Wednesday 3 May
Where: Foyer of the ABC building, South Bank
Please RSVP on humanitix
In case you missed last month’s talk by Dr Adam Snoswell – ChatGPT - Is it hype or the next big thing? – you can watch the video back here. Here’s Rosie Ryan’s useful write-up:
Are large language models like ChatGPT just "hype", or actually a step closer to a useful and general artificial intelligence? Our March event had the intriguing premise of answering this question, with a talk from computer scientist and postdoctoral researcher Aaron Snoswell.
I found it useful to think about the distinction Aaron made between the "tech moment" that's happening right now — that is, the developments in transformer architecture that are making the magic happen, so to speak — and the "cultural moment" that's happening too, where this technology has really captured the public imagination.
My notes from the talk are full of clever metaphors. A large language model is "like an advanced autocomplete", "a stochastic parrot" that mimics humans randomly with no idea of what they're saying, to be trusted "as much as you'd trust a random stranger on the internet".
(For more on the stochastic parrot metaphor, read this feature from New York magazine on linguist Emily Bender that Aaron mentioned)
Useful also was the description of what they're not: the calculators we're used to dealing with, which have an anchor to reality. They often "sound" confident and authoritative while being completely wrong. In this sense they're fundamentally different to how we usually think about machines and computers.
This last bit feels like one of the biggest challenges to me — making sure people do understand what's different about these technologies compared to what we're used to. But Aaron's not entirely pessimistic about the future. I was heartened to hear his hope that we might be prompted to cherish human-made content more, and not less.
Have you seen this?
A few more links on the generative AI beat to supplement Aaron’s excellent talk:
Nick Diakopoulos asks Can GPT-4 Help Cover Events? and presents a generated summary of a discussion panel from the AI, Media, and Democracy Lab at the University of Amsterdam.
Emily Bender argues that ChatGPT is not intelligent, speaking with Paris Marx from the podcast Tech Won’t Save Us.
Kevin Schaul, Szu Yu Chen and Nitasha Tiku look Inside the secret list of websites that make AI like ChatGPT sound smart in the Washington Post.
Nilay Patel in the Verge poses an ‘impossible legal trap’ for Google, stuck between Youtube’s need to keep copyright-holders happy and an AI division that wants to train on big datasets under the guise of fair use.
Lastly, on this month’s theme of tools, Washington Post journo Steven Rich’s newsletter Building Tables in back, combining thoughts and links on data journalism with wood-working projects from himself and others.
Looking forward to seeing you next week!