Software developer at a big library, cyclist, photographer, hiker, reader. Email: chris@improbable.org
24124 stories
·
215 followers

On having no visual memory

1 Share

I have aphantasia—no ability to create images in my mind, or to remember things in a visual way.

For well over half of my life I lived under the assumption that suggestions to visualize things were entirely metaphorical, and also that I was incredibly stupid in certain ways, finding things that everyone else seemed to find easy a challenge. I struggle to learn graphical interfaces, I’ve mixed up speakers and their talks at conferences, I spend a lot of time staring at icons on my phone—as far as I’m concerned every blue icon leads to the same app.

I store information as lists. For people I see a lot, I could probably draw a reasonably accurate picture of them, based on the list of data I’ve collected. However that information can never be as detailed as being able to recall an image could be. For example, perhaps you are someone I work with. I see you several times a week on video calls. I know lots of things about how you look, including that you wear glasses. However, your glasses were unremarkable to me and so I’ve not stored any particular information about them. You could dramatically change your style of glasses, I’d not notice. I’m not comparing a visual memory of you from a few days ago, I just know you wear glasses.

If I’m visiting a new place and I go for a run, I collect data points along the way to navigate back. I will not remember how the route looked. This means that I can be incredibly good at taking people somewhere that I’ve only been once, I’ve got turn by turn directions. However, if I am walking with someone and talking, or for some reason don’t actively collect information, I’m doomed. The other week I lost an entire car park, never mind my car, due to having been distracted by a message after leaving the car park and failed to collect any information about what it looked like.

I dream, but not in pictures. For example, I have a recurring dream where I’m back in a theatre, putting on pointe shoes. I know it’s a theatre because of the smell, I feel the roughness of the shoes as I put my feet into them, the creak of the leather sole as I roll through my foot. I can feel that dream as I write about it, but there’s no image involved.

When I learned about aphantasia about ten years ago, suddenly so many things made sense. It’s harder to remember certain types of things if you have no visual memory of them. I’m not an awful person because I didn’t remember that I’d met a person before, there was no way for them to look familiar to me. However, I also think it’s at the root of some of the things I’m really good at.

My lists of information, are closer to a relational database than just a set of lists. It’s no surprise to me now that I always enjoyed working with databases and could design a complex schema without needing to sketch out a diagram. I’m constantly making connections between these bits of information. Many of these connections are just amusing to me, but other times they bring up interesting paths to investigate.

I context switch very easily, I can jump between these information sets without losing my train of thought.

I can write entire articles, documents, or conference talks in my head while out for a run. It’s usually quicker for me to create content in this way than sit at a computer and think about it. I can come in from a run and type out 2,000 words, transferring what I’ve written in my head to the document.

Having discovered this about myself, I’ve found plenty of other people who experience the world in the same way, probably unsurprisingly as people with aphantasia tend to be drawn to computing and science. I find it fascinating that we are all experiencing the world so differently, and how that can so fundamentally impact the things we find easy, or difficult.

Read the whole story
Share this story
Delete

The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions

1 Share

The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions

By far the most detailed paper on prompt injection I've seen yet from OpenAI, published a few days ago and with six credited authors: Eric Wallace, Kai Xiao, Reimar Leike, Lilian Weng, Johannes Heidecke and Alex Beutel.

The paper notes that prompt injection mitigations which completely refuse any form of instruction in an untrusted prompt may not actually be ideal: some forms of instruction are harmless, and refusing them may provide a worse experience.

Instead, it proposes a hierarchy - where models are trained to consider if instructions from different levels conflict with or support the goals of the higher-level instructions - if they are aligned or misaligned with them.

The authors tested this idea by fine-tuning a model on top of GPT 3.5, and claim that it shows greatly improved performance against numerous prompt injection benchmarks.

As always with prompt injection, my key concern is that I don't think "improved" is good enough here. If you are facing an adversarial attacker reducing the chance that they might find an exploit just means they'll try harder until they find an attack that works.

The paper concludes with this note: "Finally, our current models are likely still vulnerable to powerful adversarial attacks. In the future, we will conduct more explicit adversarial training, and study more generally whether LLMs can be made sufficiently robust to enable high-stakes agentic applications."

Via @_akhaliq

Read the whole story
Share this story
Delete

TikTok and YouTube Shorts push misogynistic videos to young male watchers, study finds | Euronews

1 Share

A new study looked at how long it took for TikTok and YouTube Shorts to recommend misogynistic content to the accounts of young men.

ADVERTISEMENT

It takes on average 23 to 26 minutes of video watching for TikTok and YouTube Shorts to recommend toxic or misogynistic content to the accounts of young men, according to a new study.

The study from Dublin City University tracked the content recommended to 10 “sockpuppet” TikTok and YouTube Short accounts created by the researchers on new smartphones.

The accounts were all directed to show the search interests of 16 and 18-year-old boys either with regular content, like sports or video games, or to replicate those that purposely look for misogynistic content online.

The research found that it took roughly 23 minutes of video watching on both TikTok and YouTube Shorts for the algorithms to start recommending “toxic” content and 26 minutes to recommend “manfluencer” (male influencer) content across the different accounts.

The recommended videos could sometimes come as soon as after two minutes of viewing on YouTube Shorts and 10 minutes on TikTok for accounts that showed some interest in learning more about manfluencers – videos that are widely considered to be promoting alpha male and anti-feminist ideas.

“The findings of this report point to urgent and concerning issues for parents, teachers, policymakers, and society as a whole,” the report reads.

‘Monetisation of male insecurity’

The researchers watched nearly 29 hours of video over the 10 accounts to analyse the content of the videos that were being recommended.

The vast majority of content being suggested after two to three hours, or 400 video watches, was problematic or toxic, according to the researchers.

Once one manfluencer video was recommended and then watched by the young male account, it became a lot more likely to be recommended.

The report identified three major themes of these manfluencer videos: crisis narratives, like masculinity and the “nuclear family” are under threat; motivational videos that convince men that feeling emotions or depression can be emasculating; or debunked gender science videos that show concepts from evolutionary psychology that men and women are “hardwired” for different gender roles.

The study suggests there is also a link between manfluencer videos and right-wing conspiracy content: 13 per cent of all recommended content on TikTok and five per cent on YouTube for these accounts included these concepts.

“This monetisation of male insecurity not only serves to mainstream anti-feminist and anti-LGBTQ ideology, but may also function as a gateway to fringe Far-Right and other extreme worldviews,” the report reads.

One of the limitations of their study, the report continued, is the lack of transparency from social media companies about how their algorithms work.

That means they are missing critical information about how the platforms craft personalised content suggestions based on their previous viewing history.

Study not ‘reflective’ of TikTok user experience

YouTube Shorts was the platform to push the highest amount of "toxic content", at about 61 per cent compared to 34 per cent of TikTok’s recommendations.

The platform also fed manfluencer-curious accounts more toxic content than accounts that searched for "generic" topics.

YouTube did not respond to a request for comment.

TikTok said in an emailed statement that the Dublin City University report does not reflect how their user base would experience videos on their platform. The statement also noted that the sample size in the study is extremely limited, both in the number of accounts used and the amount of video viewed.

ADVERTISEMENT

They say toxic content makes up 34.7 per cent of what users would see on their feeds.

TikTok also says they do not allow hate speech or hateful discourse, like misogyny and transphobia on their platform and remove content that violates their community guidelines.

“If we become aware that any such actor may be on our platform, we will conduct a thorough review – including off-platform behaviour – which may result in an account ban,” TikTok’s community guidelines webpage reads.

The company says they do the same thing with their recommended videos in the “For You” feed, where the recommendation system will substitute less similar content into a person’s feed if they find two videos are a little too similar.

Users can also curate what they see by noting that a type of video does not interest them, by refreshing their feeds or filtering out certain keywords.

ADVERTISEMENT

TikTok is currently under two investigations under the EU's Digital Services Act dealing with “the protection of minors” and “addictive design” for the platform’s new TikTok Lite, which recently launched in France and Spain.

Last November, YouTube was asked by the European Commission to provide information on how it is protecting minors online under the new Digital Services Act.

Account-based content moderation needed

The report highlights some key recommendations for schools, parents, and social media companies.

It suggests that social media companies not only do content moderation based on the videos but also regulate what accounts are able to post.

Ireland has a new media regulator, called the Coimisiún na Meán, which the report suggests social media companies work with to "highlight illegal, harmful and borderline content".

ADVERTISEMENT

For schools, the report suggests promoting positive male role models in the classroom should be prioritised as a way to “promote an educative rather than punitive response to boys’ behaviours”. It also suggests schools focus on promoting critical digital literacy skills.

Parents could also open discussions with their teenage boys to understand why they idealise certain influencers and encourage them to “engage with relatable resources”.

TikTok said in their statement that they have parental resources to help with those conversations.

Read the whole story
Share this story
Delete

AI Is Poisoning Reddit to Promote Products and Game Google With 'Parasite SEO'

1 Share

Advertisement

A market for manipulating Reddit using AI have emerged.

Read the whole story
Share this story
Delete

War in Ukraine: how a French company continues to equip Putin's fleet of VIP aircraft - Le Parisien

1 Share

Read the whole story
Share this story
Delete

Opinion | Stop recycling plastic this Earth Day - The Washington Post

1 Share

Eve O. Schaub is the author of “Year of No Garbage: Recycling Lies, Plastic Problems, and One Woman’s Trashy Journey to Zero Waste.”

The time has come for us to stop “recycling” plastic. Plastic as a material is not recyclable, and the very best thing we can do to celebrate Earth Day this year is to acknowledge that fact.

This seems counterintuitive, I know. We’ve been told for decades that the answer to the plastic-waste crisis is more, better recycling: If only we sorted better! If only we had better access to recycling technologies! If only we washed and dried our plastics more adequately! This is all a smokescreen, designed to distract us from the truth that plastic recycling — if by “recycling” we mean converting a used material into a new material of similar value and function — is a myth.

Unlike paper, glass and metal, plastic is not easily, efficiently turned into new products. What passes for “recycling” plastic is costly, energy-intensive and toxic. On top of all that, the process requires the addition of a shocking amount of new virgin plastic — around 70 percent — to hold the newly formed plastic item together. As a result, only about 5 percent of plastic gets “recycled” (or, more accurately, “downcycled” into a product of inferior quality). Compare that with a 68 percent recycling rate for paper and cardboard.

Considering that, as a society, we’ve been actively trying to get better at plastic recycling since the 1970s, 5 percent represents a colossal, unequivocal failure. It tells us that plastic “recycling” is, at heart, an empty, performative gesture.

Many environmentalists will protest this assertion. They might correctly point out that plastics labeled with the resin identification code of 1 or 2 (the number inside the “chasing arrows” triangle on many plastics) have a higher measure of recycling success: about 30 percent. Shouldn’t we support recycling at least this plastic?

For a long time, I thought so.

But this brings us to another myth: that plastic is harmless to human health. What many people do not know is that plastic is made from two ingredients: fossil fuels and toxic chemicals. When we say toxic chemicals, we are talking about some very bad actors: heavy metals, per- and polyfluoroalkyl substances (PFAS), flame retardants and persistent organic pollutants. Tens of thousands of proprietary chemical formulas are involved in the production of plastic, most of which have never been tested for their effects on human health, although many are known to be endocrine disruptors, fertility inhibitors and carcinogens.

What this means is that even if we were to get better at recycling plastic, we shouldn’t want to. When you grind up, melt and re-form a bunch of plastic (with the addition of lots of new virgin plastic to bind it together), all those thousands of toxic plastic chemicals combine to make a Frankenstein material that has what scientists call “non-intentionally added substances” in it. Which is to say that chemicals that are not supposed to be there start showing up. A study last year concluded that recycled plastics contain “an unknown number of chemical compounds at unknown concentrations.” In 2021, a Canadian study concluded that plastic is “not suitable for processing into food grade PCR,” referring to post-consumer resin.

The upshot? You do not want your food wrapped in recycled, mystery-ingredient plastic.

But what if we use recycled plastic only for nonfood items such as picnic benches? Then we have yet another deeply troubling aspect of plastic to deal with: microplastics. We’ve been hearing more and more about these lately, because scientists are finding them everywhere they look — in the environment and in the human body.

The chemical composition of all plastic — whatever the type — is a synthetic polymer that doesn’t break down or go away, ever. Instead, it breaks up into smaller and smaller pieces until it turns into microplastics or even nanoplastics. These tiny particles are still plastic, still toxic, but now so small that we are eating and breathing them all the time. Microplastics have been discovered in human lungs, bloodstream and breastmilk, as well as in the placenta of unborn babies. Scientists have found microplastics in sperm, testes and the brain.

The effect of all this plastic in our bodies is still being revealed, but we know it is substantial. A recent study concluded that the disease burden from plastic exposure includes preterm birth, obesity, heart disease and cancer, and the health-care cost was $249 billion in 2018 alone. The human body has become the trash can of our plastics-addicted culture.

Trying to recycle plastic makes the microplastics problem even worse. A study of just one plastics recycling facility discovered that it might be washing 3 million pounds of microplastics into its wastewater every year — all of which ends up being deposited in our city water systems or dumped into the environment.

At this very moment, we all have microplastics coursing through our bodies. This is not the fault of not enough recycling. This is the fault of too much plastic. So I say: Let’s treat plastic like the toxic waste it is and send it where it can hurt people the least.

Right now, that place is the landfill.

Then we need to get to work on the real solution: making a whole lot less of it.

Read the whole story
Share this story
Delete
Next Page of Stories