
Computer science professor @DalhousieU / @dalfcs. #SoftwareEngineering, #design, #ResearchMethods, #education, #academia.
Halifax, Canada
Joined July 2009
- Tweets2.2K
- Following836
- Followers2.3K
- Likes2.3K
Pinned Tweet
I'm taking a break from Twitter at least until this asshole stops calling the shots. I feel complicit just being here. Find me on mastodon: @paulralph@mastodon.acm.org
On Feb 20 my sister, Melissa, was struck by an SUV breaking her pelvis in 3 places, possibly ending her career in veterinary medicine. I know times are tight for many but if you can spare a few dollars some of her friends have set up a go fund me: gofundme.com/f/melissa-ralph…
AGAIN @KiaCanada with the borked #ev9 launch? Seriously? First the link to the waiting room is wrong, then I finally get in, and get the exact same problem as last time, and now I'm 2000 deep in the queue AGAIN.
Replying to @ProfPaulRalph @KiaCanada
There were no prices. Just the option to pick 1 of 4 trims, and the exterior/interior colour options. No prices
I've been to the front of @KiaCanada's online queue to order an EV9 three times now and the system's bugged out each time and kicked me to the back of the queue. Now it says there's 50,000 people behind me. Riiiiight.
For @ConfEASE’24 I humbly request that @ProfPaulRalph & Dag Sjøberg run an interactive session:
people bring their papers, the threats to validity section (& construct validity) are discussed in small groups (maybe using a checklist?), and improvements are shared with everyone.
"Continuing with traditional peer review is scientifically preposterous and morally unconscionable" —@profpaulralph at #ease2023
😲🤯☠️
Fascinating stats by @ProfPaulRalph on reviewer reliability and the experiment run at @ConfEASE’23! 📉 #ease2023
Fascinating stats by @ProfPaulRalph on reviewer reliability and the experiment run at @ConfEASE’23! 📉 #ease2023
'How can you discuss construct validity if you haven't defined any constructs?' - paraphrasing Dag Sjøberg presenting Improving the Reporting of Threats to Construct Validity @ #EASE2023
Looking forward to @ConfEASE'23, starting tomorrow! I'll be presenting our work on Boundary Objects in DevOps, co-authored with @R_C_Heinrich & @RebekkaWohlrab.
Come join the Development Processes Session, chaired by @eraytuzun! 🧑💻💬 @HPI_EPIC @HPI_DE
conf.researchr.org/details/e…
Screen sharing with a student today:
S: "I don't understand this function.."
Me: "<Explains>...change line 11"
S: "<Starts typing...>
Copilot: "<Offers suggestion, no explanation>"
S: "<Immediately accepts suggestion without reading, still doesn't understand, code is wrong>"
AI class project suggestion: make a deepfake sextape featuring every member of parliament and send it to their offices. Tell them they have 30 days to address this AI shitshow or you'll release it across "the dark web." 🙃
The FBI released a warning on deepfakes for sextortion attacks - blackmailing someone with a fake sexual photo of you/your family member and being asked to pay up to take it down.
www-bleepingcomputer-com.cdn…
Now with AI powered tools, bad actors can generate thousands of such content daily. Even if a small % of victims pay up, there is enough financial incentive for the bad actor to conduct this attack.
Using fake images for sextortion is the same as a computer ransomware, which will lock any computer it gets its hands on and demand money/btc to get back the file. The more computers the ransomware can reach, the more money it makes.
The most common counter-argument to this is this is no different than using Photoshop.
However, I argue that the economics behind AI-powered fakes are indeed is completely different from using Photoshop:
(1) Volume: AI digital tools allow manipulation and sextortion at a scale that wasn't possible before.
It's a numbers game - the more people a bad actor can target, the higher their chance of success and ranson.
Yes, bad actors created fake images before such tools existed, but now, they can directly target an incredibly high number of potential victims.
(2) Speed: The time taken to create such fakes is reducing. With photoshop, it would take a bad actor hours to create a blackmail photo. Now it takes minutes, and it can even be competely automated.
(3) Access: Photoshop requires some level of skill. With these tools and more coming, anyone can do it.
What do you think?
The survey @macleans uses to rank universities is so stupid and unscientific I couldn't even finish filling it out.
I always thought of a "climate refugee" as a person immigrating to my area. I never thought I'd *be* a climate refugee. And yet here we are.
People who think software is badly designed... consider suburbs. Many suburbs are built with a single entrance/exit and no firefighting infrastructure. Now some of them are on fire and I'm living in a dorm until further notice.
Halifax Fire had warned that 3 subdivisions evacuated in Tantallon fire were built with ‘inadequate water sources to fight fires’ halifaxexaminer.ca/governmen… by me
This is actually how "confidence intervals" work in statistics. I'm being 100% serious. Read on. This is a really fun concept.
xitter.cf/HumansNoContext/status…


















