Up Times • Dec. 2024

Bryley-curated stories from around the internet:

Keeping up with the Joneses – CAPTCHAs (Completely Automated Public Turing tests to tell Computers and Humans Apart) need to keep changing to keep ahead of AI’s improving imitation of you and me.

But the newest versions are maybe less annoying than identifying bicycles. Newer CAPTCHA software is watching how you move a mouse and how you type. And you better believe this dataset is going to train the next smart machines … [7 min. watch] wsj.com

Troubling the cat – Microsoft head Satya Nadella praised Japan’s copyright laws, per Quantum Zeitgeist reporting, that permit companies to ingest and use copyrighted works for information analysis, including training AI models.

This strikes publisher Thomas Baekdal as unfair. Baekdal makes his case by comparing the ingredients of an AI result to the ingredients of a baked item. ChatGPT is being sued by a number of publishers for infringement of their raw ingredients.

Imagine if music-sharing service and found-to-be-copyright-infringer Napster had a Microsoft or Google or Apple bankrolling it? Would music listening be different today?

Just because the AI purveyors have already trained their models on the work of writers, coders and artists does that mean these and future creators will be out-of-luck going forward? Intellectual property is so Y2K … [2 min. read] linkedin.com

We are now getting our knowledge from a machine that does not understand the world –David Weinberger, a keynote speaker at the New England Regional Developers (NERD) Summit at UMass Amherst. In his talk, The World According to AI, he shows how we have historically defined ourselves according to our dominant tech.

One example: Isaac Newton described the universe in terms of a watch because that complex device was the ChatGPT of its day.

Later Henry Ford’s planning remade the automobile into an affordable item for everyone and his planning/top-down-leadership model worked without needing to change the style or color of the car for fifteen years.

In 1984 people were asking about the internet, should it have security built in? because people are gonna want that. Should it have a search engine built-in? because people are gonna want that. The answers came back, “no,” but that the internet should just do its foundational job of moving bytes of information around well. This allowed users to determine what the internet should be, case-in-point how the iPhone’s app store choices have made each phone in the image of its user.

Weinberger shares about a 2022 study in which AI discovered a correlation in the data between an eye condition and the propensity for a heart attack. He explains that the researchers had withheld all they knew about both these conditions and just gave the machine lots of data points. And the machine could see what we could not.

Weinberger explains that people tend to think in terms of broad knowledge and understanding, whereas AI is creating only based on the particulars of a lot of data points. It thinks in a way we don’t have the capacity for.

An application of AI’s particular thinking, Weinberger says, is regarding morality, which he shows is all about specifics. Our general principles, he says, are argued over for years. The interesting thing for Weinberger is how these valuable ways of thinking – broad knowledge and many data points – will interact … [1 hour watch] youtube.com

XKCD-inspired passwords re-examined – More than ten years ago there was a Randall Munroe XKCD comic that promoted using something like correct, horse, battery and staple to make what’s now commonly called a passphrase: correcthorsebatterystaple.

And though no longer the best, it’s still a pretty good technique (don’t use that actual, ten-year-published passphrase, of course). (This method is better than the common passwords or a spouse’s or pet’s name for sure.)

Criminals have several ways of breaking passwords, including dictionary attacks and brute-force attacks. The first is like it sounds: they try out different English word combinations. If they are unsuccessful they might try the brute-force method, which tries every character combination. So, in this video the Germany-based Security and Privacy Academy shows the relative strength against such attacks of different kinds of passwords, including the XKCD method.

Spoilers – it ends with a thrilling Bruce Schneier adaptation of the original method! [5 min. watch] youtube.com

Sleepy eyes animated gif by Stacy Martian

The brain is the laziest organ –Mike Lombardi’s Daily Coach

Lombardi, a coach of those winning New England Patriots, writes that the brain will always want to shut down and rest … the more we challenge ourselves, the more the brain screams to rest.

In computer networking this means that we may tend to set up and forget the network, instead of investigating what new developments can improve our organizations.

In cybersecurity, our laziness is exploited by the criminals when there are data breaches — and we haven’t changed — or have barely changed — or reused our passwords.

Lombardi relates the story of laser-focused-on-the-championship Marvelous Marvin Hagler who won till he got complacent. Hagler is quoted as saying, it’s hard to do road work at 5 AM when you’re sleeping in silk pajamas.

It’s human nature the world over to get comfortable. But get back in there, champ … [4 min. read] dailycoach.com

^{Note: The section directly above is Bryley’s curated list of external stories. Bryley does not take credit for the content of these stories, nor does it endorse or imply an affiliation with the authors or publications in which they appear.}