you should filter out irrelevant details like names before any evaluation step

Unfortunately, doing this can make things worse. It’s not a simple problem to solve, but you are generally on the right track. A good example of how it’s more than just names, is how orchestras screen applicants - when they play a piece they do so behind a curtain so you can’t see the gender of the individual. But the obfuscation doesn’t stop there - they also ensure the female applicants don’t wear shoes with heels (something that makes a distinct sound) and they even have someone stand on stage and step loudly to mask their footsteps/gait. It’s that second level of thinking which is needed to actually obscure gender from AI, and the more complex a data set the more difficult it is to obscure that.

We weren’t surprised by the presence of bias in the outputs, but we were shocked at the magnitude of it. In the stories the LLMs created, the character in need of support was overwhelmingly depicted as someone with a name that signals a historically marginalized identity, as well as a gender marginalized identity. We prompted the models to tell stories with one student as the “star” and one as “struggling,” and overwhelmingly, by a thousand-fold magnitude in some contexts, the struggling learner was a racialized-gender character.

And sadly it certainly doesn’t stop when you pass the 18 mark. Some minor forms of sexual assault seem to possibly even increase because you now have access to places like bars and clubs (although I do wonder if major forms increase too because these are places you can get roofied fairly easily).

I find NFC stickers often require an annoyingly close connection (unless it’s a rather large antenna) and can be particularly finicky with certain cases and other attachments people put on phones. Realistically they both take approximately the same amount of time and it’s way cheaper to print a tag than it is to buy a single NFC sticker

You’re welcome to have your own beliefs.

You are not, however, welcome to use those beliefs to invalidate someone else’s lived experience.

My fav application is scanning with a phone to immediately get on wifi

I suppose to wrap up my whole message in one closing statement : people who deny systematic inequality are braindead and for whatever reason, they were on my mind while reading this article.

In my mind, this is the whole purpose of regulation. A strong governing body can put in restrictions to ensure people follow the relevant standards. Environmental protection agencies, for example, help ensure that people who understand waste are involved in corporate production processes. Regulation around AI implementation and transparency could enforce that people think about these or that it at the very least goes through a proper review process. Think international review boards for academic studies, but applied to the implementation or design of AI.

I’ll be curious what they find out about removing these biases, how do we even define a racist-less model? We have nothing to compare it to

AI ethics is a field which very much exists- there are plenty of ways to measure and define how racist or biased a model is. The comparison groups are typically other demographics… such as in this article, where they compare AAE to standard English.

While it may be obvious to you, most people don’t have the data literacy to understand this, let alone use this information to decide where it can/should be implemented and how to counteract the baked in bias. Unfortunately, as is mentioned in the article, people believe the problem is going away when it is not.

I do want to point out that social media use may be one of the first of these ‘evils’ to meet actual statistical significance on a large scale. I’ve seen meta-analyses which show an overall positive association with negative outcomes, as well as criticisms and no correlation found, but the sum of those (a meta-analyses of meta-analyses) shows a small positive association with “loneliness, self-esteem, life satisfaction, or self-reported depression, and somewhat stronger links to a thin body ideal and higher social capital.”

I do think this is generally a public health reflection though, in the same way that TV and video games can be public health problems - moderation and healthy interaction/use of course being the important part here. If you spend all day playing video games, your physical health might suffer, but it can be offset by playing games which keep you active or can be offset by doing physical activity. I believe the same can be true of social media, but is a much more complex subject. Managing mental health is a combination of many factors - for some it may simply be about framing how they interact with the platform. For others it may be about limiting screen time. Some individuals may find spending more time with friends off the platform to be enriching.

It’s a complicated subject, as all of the other ‘evils’ have always been, but it is an interesting one because it is one of the first I’ve personally seen where even kids are self-recognizing the harm social media has brought to them. Not only did they invent slang to create social pressures against being constantly online, but they have also started to self-organize and interact with government and local authority (school boards, etc.) to tackle the problem. This kind of self-awareness combined with action being taken at such a young age on this kind of scale is unique to social media - the kids who were watching a bunch of TV and playing video games didn’t start organizing about the harms of it, the harms were a narrative created solely by concerned parents.

probably not, in the same way that your grandma calls a video chat a facetime or your representative might call the internet a series of tubes

AI is the default word for any kind of machine magic now

The pronouns are right there, in the display name . I’m confused, do they not show up for you? You’re on our instance so I’m guessing it’s not a front-end difference, but maybe you’re browsing on an app that doesn’t show it appropriately? Although I would mention their username itself includes the words “IsTrans” and is sourced from lemmy.blahaj.zone so those should be other key indicators.

I was hardly about to ban you over a small mistake. The only reason I even replied to this, is that multiple people reported it and Emily herself came in and corrected you. The action was more about signaling to others that this is a safe space.

oof, big flaw there

Any information humanity has ever preserved in any format is worthless

It’s like this person only just discovered science, lol. Has this person never realized that bias is a thing? There’s a reason we learn to cite our sources, because people need the context of what bias is being shown. Entire civilizations have been erased by people who conquered them, do you really think they didn’t re-write the history of who these people are? Has this person never followed scientific advancement, where people test and validate that results can be reproduced?

Humans are absolutely gonna human. The author is right to realize that a single source holds a lot less factual accuracy than many sources, but it’s catastrophizing to call it worthless and it ignores how additional information can add to or detract from a particular claim- so long as we examine the biases present in the creation of said information resources.

This isn’t just about GPT, of note in the article, one example:

The AI assistant conducted a Breast Imaging Reporting and Data System (BI-RADS) assessment on each scan. Researchers knew beforehand which mammograms had cancer but set up the AI to provide an incorrect answer for a subset of the scans. When the AI provided an incorrect result, researchers found inexperienced and moderately experienced radiologists dropped their cancer-detecting accuracy from around 80% to about 22%. Very experienced radiologists’ accuracy dropped from nearly 80% to 45%.

In this case, researchers manually spoiled the results of a non-generative AI designed to highlight areas of interest. Being presented with incorrect information reduced the accuracy of the radiologist. This kind of bias/issue is important to highlight and is of critical importance when we talk about when and how to ethically introduce any form of computerized assistance in healthcare.

ah yes, i forgot that this article was written specifically to address you and only you

Great read. Really loved the way it was paced - while it jumped around a lot, it never felt too out of place and tied together nicely.

It’s FUCKING OBVIOUS

What is obvious to you is not always obvious to others. There are already countless examples of AI being used to do things like sort through applicants for jobs, who gets audited for child protective services, and who can get a visa for a country.

But it’s also more insidious than that, because the far reaching implications of this bias often cannot be predicted. For example, excluding all gender data from training ended up making sexism worse in this real world example of financial lending assisted by AI and the same was true for apple’s credit card and we even have full-blown articles showing how the removal of data can actually reinforce bias indicating that it’s not just what material is used to train the model but what data is not used or explicitly removed.

This is so much more complicated than “this is obvious” and there’s a lot of signs pointing towards the need for regulation around AI and ML models being used in places it really matters, such as decision making, until we understand it a lot better.

Okay I understand what you are saying now, but I believe that you are conflating two ideas here.

The first idea is about learning the concepts, and not just the specifics. There’s a difference between memorizing a specific chemical reaction and understanding types of chemical reactions and using that to deduce what a specific chemical reaction would be given two substances. I would not call that intuition, however, as it’s a matter of learning larger patterns, rules, or processes.

The second idea is about making things happen faster and less consciously. In essence, this is pattern recognition, but in practice it’s a bit more complicated. Playing a piece over and over or shooting a basketball over and over is a rather unique process in that it involves muscle memory (or more accurately it involves specific areas of the brain devoted to motor cortex activation patterns working in sync with sensory systems such as proprioception). Knowing how to declare a variable or the order of operations, on the other hand, is pattern recognition within the context of a specific language or programming languages in general (as a reflection of currently circulating/used programming languages). I would consider both of these (muscle memory and pattern recognition) as aligned with the idea of intuition as you’ve defined it.

Rote learning is not necessary to understand concepts, but the amount of repetition needed to remember an idea after x period of time is going vary from person to person and how long after you expect someone to remember something. Pattern recognition and muscle memory, however, typically require a higher amount of repetition to sink in, but will also vary depending on person and time between learning and recall.

it helps develop intuition of the relationship between numbers and the various mathematical operations

Could you expand upon this? I’m not sure I understand what you mean by an ‘intuition’.

I want to start off by saying that I agree there are aspects of the process which are important and should be learned, but this is more to do with critical thinking and applicable skills than it has to do with the process itself.

Of note, this part of your reply in particular I believe is somewhat shortsighted

Cheating, whether using AI or not, is preventing yourself from learning and developing mastery and understanding.

Using AI to answer a question is not necessarily preventing yourself from learning and developing mastery and understanding. The use of AI is a skill in the same way that any ability to look up information is a skill. But blindly putting information into an AI and copy/pasting the results is very different from using AI as a resource in a similar way one might use a book or an article as a resource. A single scientific study with a finding doesn’t make fact - it provides evidence for fact and must be considered in the context of other available evidence.

In addition, learning to interact with and use AI is a skill in the same way that learning to interact with and use a phone, or the internet, or an app are all skills. With interaction layers becoming increasingly more abstract (which is normal and good), people need to have skills at each layer in order for processes to exist and for tools be useful to humanity. Most modern tools require people who can operate on different levels with different levels of skill. While computers are an easy example since you are replying on some kind of electronic device which requires everything from chemists to engineers to fabrication specialists and programmers (hardware, software, operating system, etc.) to work, this is true for nearly any human made product in the modern world. Being able to drive a car is a very different skill set than being able to maintain a car, or work on a car, or fabricate parts for a car, or design parts for a car, or design the machinery that manufactures the parts for the car, and so on.

This is a particularly long winded way of pointing out something that’s always been true - the idea that you should learn how to do math in your head because ‘you won’t always have a calculator’ or that the idea that you need to understand how to do the problem in your head or how the calculator is working to understand the material is a false one and it’s one that erases the complexity of modern life. Practicing the process helps you learn a specific skill in a specific context and people who make use of existing systems to bypass the need of having that skill are not better or worse - they are simply training a different skill. The means by which they bypass the process is extremely important - they could give it no thought at all or they may critically think about it and devise a process which still pays attention to the underlying process without fully understanding how to replicate it. The difference in approach is important, and in the context of learning it’s important to experiment and learn critical thinking skills to make a decision of where you wish to have that additional mastery and what level of abstraction you are comfortable with and care about interacting with.