What does AI actually do for us?
![[Post New]](/jforum/templates/default/images/icon_minipost_new.gif){kind=icon}
Anonymous
But future generations wont be able to because youve been able to create and organize thoughts without AI you have your own INTELLIGENCE versus having to buy it from overlords in 10-20 years. Thanks so much for your service! |
Anonymous
I work at a university and most departments have an administrator who makes a big deal every summer about updating the textbook orders -- making sure the syllabi and orders at the bookstore link to the latest edition of the textbook, making sure faculty know there is a new edition available, etc. Presumably there's also someone at the bookstore who makes a big deal out of this every summer too. Last week I was sent an excel spreadsheet and instructed to (as the instructor) make sure the edition listed was correct, and otherwise to update it to the correct ISBN, etc Instead of going to the publisher's website and checking all the stuff, I fed the whole darned spreadsheet to the AI, said "check that all of these textbooks have the correct edition, and if not, give the new ISBN, etc." Basically, the AI did the administrator's entire job for the month of June in like 10 minutes. As the administrator I would have then asked it to email all the faculty whose textbook has changed and inform them, etc. I have the feeling that the administrator whose job this currently is is NOT going to let anyone know you can feed the whole thing to an AI and do your month's long job in ten minutes. I'm sure this is true in many enterprises at the moment. |
Anonymous
This. OpenEvidence is actually excellent at coming up with differential diagnoses for a long nonsensical list of symptoms that a patient comes in with. It helps me check my bias. I will see a patient, listen to them, then do my physical exam and ask some pertinent follow up questions, and I've had an idea as to what the problem is from the first 2 minutes of the visit usually. And I'm still usually right. But I do often plug it all into OpenEvidence at the end of the day if a patient seemed a little complex or unusual, and see what they come up with. Sometimes they point me towards ideas that I hadn't truly considered, and it has occasionally made me change my plan of care to include workup for some additional stuff. It's also very easy just from a quick fact check perspective to ask it "what's the current treatment recommendation for Early Disseminated Lyme for a 6 year old" because it spits out the Red Book recommendation without me having to go and find the Red Book, or without me having to log into UpToDate and search through the article on Lyme treatment. It gives the same information, from reliable medical sources (like UpToDate does), but it takes me 10 seconds as opposed to 120 seconds. This seems stupid to anyone who isn't a physican/NP/PA in urgent care/ primary care/ fast track ER, but anyone who is, knows exactly how nice it is to shave 2 minutes off of a patient encounter. |
Anonymous
That is a result of your employer's policy, though, not because it is not possible to make secure and confidential closed AI systems to input the information. If you trust your banking data and your health data to be online (or accept that it is online, at least), then there is no good reason why legal materials can't be online too. It is only a matter of time. |
Anonymous
I feel like lawyers are the first thing AI could replace. |
Anonymous
|
My primary care doctor asks at the beginning of the appointment if it’s okay to record our conversation. If I agree, AI can generate a summary of the visit and clinical notes.
I asked her what she thought of this AI use and she really likes it. She doesn’t have to type nearly as much during the visit (so she can be more present) and just needs to edit what AI generates. PCPs have a tough job (I threw several different concerns at her during a follow up appointment) and she is great, so I’m glad it makes things a bit easier for her. That being said, I have noticed errors in the written notes from other doctors who are also using AI so it’s not perfect. |
Anonymous
| Our veterinarian now uses the AI note taking thing too. it's not just for humans! |
Anonymous
Definitely not. It can help with process issues and save quite a bit of time but AI is notorious for hallucinating cases, statutes, not making the right connections from the cases it does correctly cite. |
Anonymous
Maybe. There's still an open question about how Rule 6(e) applies here. You're oversimplifying, likely because you're not familiar with grand jury practice. |
Anonymous
You do have a clear hostility towards AI. You should think about why you're so resistant and fearful. This isn't about replacing my brain with blind faith in something else, any more than I had faith a book was accurate enough. I use AI as an intelligent tool to engage with and I learn from it. That's why I find it exciting, I'm constantly learning new things from the AI, whether a new program to help me do my work faster or how to repair the ice maker in my fridge or the underlying causes attracting certain people to certain politicians. It's interesting. AI is the logical progression from the google search engine. But instead of having to filter through 20 links to find nuggets in each one that would help me find what I'm looking for, it synthesizes all of them into one direct output. It sifts through vast amount of information faster and pulls out what is relevant. It gives me more time to do other things. And the speed stimulates creativity in me instead of being bogged down in the process of research. It opens up new capacities in understanding things. That's why I've come to love it. But what you get out of the AI will be directly related to the effort you put into it and certainly you need to develop your tools and strategies for working with the AI and understanding its limitations and flaws and how to work around it or adjust for it. |
Anonymous
| AI drafted the majority of my EBay listing for my son's Pokemon card so I didn't need to use any thought completing that task. Well, not as much thought. I did have to review the draft and delete false information about the card being autographed by Will.I.Am but hey. |
Anonymous
Yes I do because its being implemented without any guardrails and those in charge of it have taken over our political system so theres no pushback. And AI is not perfect and its being presented as a final source without flaw. Its also xenophobic, perpetuates racial and economic issues, etc. Using it as a final source is the flaw in your approach. It doesnt seem to be in addition to approach because of the bolded. If its deciding what is relevant then who is doing the thinking? |
Anonymous
| All of these long responses are AI right? Lots and lots of words with no real point. |
Anonymous
There's plenty of guardrails and complaints about safeytism in AI. And it is a constant topic everywhere. The biggest challenge is an AI that takes decisions into its own hands, and that can be a problem and that is where the real debates are happening. The rest of your post is your bias and speaks more about you than AI. How is AI racist? Or is it summarzing information you don't like, therefore it is racist? AI will confirm that black men commit a disproportionate amount of violence in this country. Is that racist or true? Because it is true. And what AI will also do is to give the reasons for why this might be the case, which will also talk about a legacy of institutional racism. So people who cry racism are only upset that AI isn't automatically confirming their biases. Your fears are cliches and it is amusing what it says about certain mindsets among the progressive left who fear what they can't control easily so you trot out the usual list of litanies that's applied to everything you don't like. Racism! Bigotry! Xenophobia! |
Anonymous
Oh f6ck I thought I was communicating with someone living in reality. My bad. https://www.ohchr.org/en/stories/2024/07/racism-and-ai-bias-past-leads-bias-future https://hai.stanford.edu/news/covert-racism-ai-how-language-models-are-reinforcing-outdated-stereotypes https://link.springer.com/article/10.1007/s00146-024-01893-4 But since youre so in love with AI heres what Claude said when I asked if AI perpetuates racism and xenophobia "Yes — there is strong, well‑documented evidence that AI systems can perpetuate racism and xenophobia, and this has been confirmed by academic research, UN human‑rights investigations, and real‑world audits of deployed AI systems. The core finding across these sources is that AI reproduces and amplifies the biases present in its training data, design choices, and deployment context, creating discriminatory outcomes even when race or ethnicity are not explicitly provided as inputs. 📌 Key Evidence That AI Perpetuates Racism and Xenophobia 1. UN Human Rights Investigations The UN Special Rapporteur on racism reports that the assumption that AI is “neutral” is false — AI systems can perpetuate racial discrimination through biased data, biased algorithms, and biased deployment. Examples include predictive policing systems that reinforce historical over‑policing of minority neighborhoods, creating feedback loops where “bias from the past leads to bias in the future.” 2. Stanford Research on Language Models A 2024 Stanford study found that major language models (including GPT‑3.5, GPT‑4, RoBERTa, and T5) generate covertly racist outputs, especially against speakers of African American English (AAE). Models assigned AAE speakers: lower‑prestige jobs higher likelihood of criminal guilt harsher sentences (e.g., death penalty over life imprisonment) This occurred even when race was not mentioned, showing that dialect alone triggered discriminatory behavior. 3. Predictive Policing and Criminal Justice Predictive policing algorithms use historical arrest data — which is already racially skewed — and therefore direct more police presence to minority neighborhoods, reinforcing racial disparities. This creates a self‑reinforcing cycle where over‑policed communities generate more “evidence” of crime, which the algorithm then interprets as justification for further policing. 4. Systemic Bias in AI Design and Data A UN report identifies four ways AI contributes to racial discrimination: Biased or incomplete data Algorithm design choices Intentional discriminatory use Lack of accountability mechanisms These issues appear across healthcare, employment, policing, and immigration systems. 5. Academic Reviews of AI Bias Scholarly analyses show that AI systems suffer from: Input bias (biased or unrepresentative training data) System bias (biases introduced during model design) Application bias (biases arising from how AI is used in the real world) These biases lead to injustice, discrimination, and harmful outcomes, especially for racial and ethnic minorities. 📌 What This Means in Practice Across domains, biased AI has produced: Higher misidentification rates for darker‑skinned individuals in facial recognition Unequal healthcare predictions for Black patients Discriminatory hiring recommendations Xenophobic patterns in immigration risk scoring and border‑control algorithms Racialized language outputs from large language models These are not isolated incidents — they reflect structural patterns. 📌 Why This Happens AI systems learn from human‑generated data. When that data reflects: historical racism unequal policing biased hiring discriminatory language xenophobic narratives …the AI absorbs and reproduces those patterns, often at scale and with a veneer of objectivity. 📌 If you want, I can go deeper into any of these areas: predictive policing language‑model racism bias in facial recognition" |