Kickoff For May 5, 2024
Welcome to another week. Let's get it started with these links:
AI Mistakes Are Very Different Than Human Mistakes — Wherein we get a look at the kinds of errors (both major and minor) that large language models and AI chatbot spit out, why those errors can be damaging, and get some advice about how to deal with them.
From the article:
In some cases, what’s bizarre about LLMs is that they act more like humans than we think they should. For example, some researchers have tested the hypothesis that LLMs perform better when offered a cash reward or threatened with death.
‘The internet hasn’t made us bad, we were already like that’: The mistake of yearning for the ‘friendly’ online world of 20 years ago — Wherein we learn about the problems with online nostalgia, and how much of what we're encountering in the online world today (good and bad) has always existed in that sphere.
From the article:
Like all nostalgic escapism, the myth about a world wide web before the age of sarcasm (and the dominance of big companies) where everything was more sincere and simpler is a melancholic trap. The Austrian poet Ingeborg Bachmann wrote that when you turn 30 you discover the ability to remember, and those who were teenagers when broadband was installed in most homes are now that age. That is why the internet has filled up with memories of itself, although, with some effort, it is still possible to find new things full of a collaborative spirit.
Your brain is lying to you about the “good old days” — Wherein we learn about the power of nostalgia, how a lot of that might be false or misleading, and why all those yesterdays might not be better than the present moment.
From the article:
[S]aying the past is better than the present means making a judgment of what we mean when we say “the past.” Not everything has improved, and sometimes periods of progress are followed by periods of decline. The arc of history doesn’t only go up and to the right. But if you step back a bit, although you’ll see some dips, the trend lines are quite clear.
Fake Reviews Have Become the Internet’s Perfect Crime — Wherein Timothy Caulfield explores the world of fraudulent ratings on various websites, why they've become integral to both sellers and buyers, how difficult it's become to identify fake reviews, and how regulators (an others) are trying to fight back.
From the article:
The incentive for companies to generate positive reviews is intense. On a global level, trillions of dollars are in play. Almost the entire consumer economy has a horse in this race. Good reviews translate into profits. Alas, studies have consistently shown that generating fake or “incentivized” positive reviews works.
What an Insomniac Knows — Wherein Adam Gopnik explores the nature of sleep, consciousness, and the nasty effects that not sleeping has on our physical and mental health.
From the article:
Insomniacs seldom just get up, work for an hour, enjoy the silence of the house. This implies a state of serenity that’s exactly what we don’t have; if we could be that calm, we’d be asleep. No, we are inclined to seek out sleep in the same oscillating stages that sleep itself presents, even if that means walking fretfully, or listening to podcasts on early Christian history, or watching late-night television, searching out things that will be sufficiently distracting to keep us from dwelling on the fact that we are not sleeping without being so agitating as to keep us up even more.
Saving One Screen At A Time — Wherein Ernie Smith ponders the humble screen saver: its origins, the challenges of creating one, and how the screen saver entered into popular computing culture and use.
From the article:
The fact is, computers are designed to be much more static in their presentation of data than television-based mediums. A spreadsheet is not a video game, and odds are that spreadsheets, without interaction, could live on the screen for months. And that created burn-in challenges that, at the time spreadsheets first came on the scene, were genuine issues.