Dont need more than 8gb. It'll be enough power. IT can do audio to audio.
672 karma · joined March 23, 2022
Dont need more than 8gb. It'll be enough power. IT can do audio to audio.
But at the same time, I wonder how much of online political speech is actually poisoned by angsty immature low-info teens.
But I do expect it's less than dishonesty. People who are parodies pushing negative stereotypes or dishonest positions of their political opponents.
What would online political speech look like if everyone genuinely represented their own views as well as they could; including admitting when they simply dont know. Might be an enlightenment event on its own.
The current SP500 pe ratio is ~31
It's not unreasonable to expect that this might even be below the new average whatever that might be. In fact, the antistock market like gold/silver/bitcoin is also probably low; not high.
Even if 17 is still the correct number, 31 isnt even a bubble by definition.
This isnt adhd. It's an awful lot like adhd but isnt unique or something that would be a diagnosis.
At most you're diagnosing high curiosity and obviously openness. Just a personality type. Common of people who were exposed to the internet since childhood. You're probably dating yourself as a millennial; late 30s?
>I think there is a founder/ADHD thing. Paul Graham thinks so. Maybe even a tech person angle. What have other people experienced?
On either side of this personality traits. You have the thinkers who make connections and invent new things. So they are great at startups and entrepreneurship. But the moment the business gets beyond this step. They fall away.
The opposite personality is one who doesnt make new connections or invents new things, but knows how to execute and follow through.
Ive seen it happen in tech and politics many times now. It's something Elon musk learnt in university. You have to give up your baby to be raised by someone who can raise it right.
>And how do others cope? I don't really know this world. I do know that my old boss once called me a "flagitating laser beam". I think he meant distracted. I use a bunch of systems to cope. For a long time lists, and then Asana. Asana ruled my life. I just built my own thing to capture tasks, projects, but also knowlegde. Not sure if it will help we will see.
a couple paragraphs on the internet will never get you a diagnosis of adhd... but there are tests online that might guide. I was a huge trello fan until atlassian ruined it. Ive been thinking of coding a trello-like into my home django project.
>- Who else feels this way? - How do you manage? - Oh and how do you switch off? That is hard
Why switch it off? There is training of course; you need to learn when to listen to your guardian angel. mindful meditation will help. Go buy a muse EEG. Do get the sleep one even though its more expensive.
But no 'switch off' it's about keeping it, but being able to tell your brain not now.
What you need is an assistant or partner. Luckily for you, if you cant afford someone, you can use AI.
When someone uses an em dash, it implies they arent using a normal keyboard; not even dvorak.
Grok4 and Gemini 3 Pro top models are around the 125-130IQ range. They are rapidly moving towards ASI.
Then a large amount of $, time, and effort was spent on women to solve this problem, but since they refused to understand the cause. It really was just a boost to women. This shifts men behind. Government intentionally caused this.
https://www.nytimes.com/2025/05/13/upshot/boys-falling-behin...
Then men needed a role model. Men would need to work extra hard to just come up to parity. But government didnt like the messaging of those role models, so they censored their speech and deplatformed them.
Creating the gap/shortage in role models; but men moved to beyond their control. This new role model became huge. He'd talk for hours upon hours with anyone who would listen.
The people who wanted to censor/deplatform but couldnt. So they publicly killed him.
Your customers are the patent lawyers who need to do that research.
Your competitor is every AI.
Major security breaches are going unnoticed.
What I think is happening there. A model might only reasonably have 65,000 context. You quickly use it all up and it doesnt tell you it's truncating context, but it defaults in the center.
Eventually your chat is silently removing context and the message you think you're sending the LLM is X, Y, Z and what it's processing is X, Z. Which gives a different answer than the missing Y context.
My thought, use a new chat regularly.
The other option, go local. It's private, ask it whatever it you want. Nobody will ever know.
Totally fine with me. Imagine all the cameras you walked by even getting to the new cameras you seemingly have problems with? If you're at an airport, people have their phones out recording all the time. It's public. I want the CBSA to be recording, databasing, post-analyzing including the ability to feed a photo into their database and know who that is. That's the border.
The big question, I'm not sure. Should the data be freedom of information act accessible? I think we side on privacy here; ban the government from sharing the ID information.
>EU rules are currently stricter and US rules allow some opt-outs for people that are uncomfortable with it.
Here in Canada, no significant rules. You can ride on an international flight fully covered except eyes to see where you're going. We recently increased privacy having everyone ride with a mask.
You can opt out of the radiation/xray scans for health and religious reasons.
The things in front of you are what happens. Want to learn guitar, dont keep it in another room or a closet. In fact, remove closet doors, make things visible. There's a balance of spending too much time organizing and removing all friction to productivity.
You can even take pictures of your battlestation and ask AI what improvements could be made.
You understand your mind has multiple sensors/states; emotions, good, bad, future, past, etc. and you need to filter these. All 'predictions' about the future are lies. All rumination about the past, you dont even remember what happened correctly, move on.
Your identity is not true; it may change drastically on short notice. Some good, some bad. You identify with any of the "I ams" like "I'm poor" or "I'm socially awkward or introvert" Maintain a written list of your identity; make sure it's positive and helpful. You're the authority of this, nobody else cares. Your identity might be
I am:
Good, moral, trying to help all people.
Love incarnate to all people.
Hard worker with integrity.
Loved and worthy.
Open to change for the better.
Ask for forgiveness rapidly, and not for permission.
Keeping my health to the maximum at all times.
The smaller the identity the easier the rules of life becomes.
From my pov, what a great change.
Anyone talking without specific location, it doesnt matter.
Anyone talking positively about something, it doesn't matter.
We now know more about those interacting in a negative way.
I had only anecdotal knowledge(I know Canadian doctors are all going to the big cities) and I looked it up before posting for who would be most impacted.
You are correct the article tried to suggest rural, but fact check false. It's easy to see why NPR did this.
NPR represents urban liberals; their readers won't like reading that their healthcare costs are about to go way up.
Bad news for New York, New Jersey, California, Maryland; or more aptly labelled democrat controlled states.
Probably the only republican state impacted will be florida, but they often buck trends like this. they'll keep their immigrant doctors.
The cases where their location does matter, there wasnt a question as to where they were from.
There were virtually no surprises from my point of view.
Pretty much all the major LLMs are politically biased. Claude is very libertarian. Whereas Bing copilot is extreme far left.
What's very interesting to me, the quadrant they are in, typically quite unsuccessful at politics and typically quite unpopular.
So following the capital: Heavy investment from government, typically military, especially during cold war era. Which also forced investment from the "enemies"
Into culture golden age where going from records to 8-tracks to cassettes to cds to dvds to digital. The same product was able to sell multiple times creating large wealth that would need to be invested.
Next, you need high education, with a tapped talent pipeline so that the talent doesnt leave. Then also a culture that's against the typical; no stability, constantly looking for a new job.
The theory here though was distilled. Hong Kong experiment compounded with their 8 free trade zones and the neighboring ones like Shenzhen. Followed by many African ones, loads in UAE. Several across europe. Canada even has several now. Whereas the USA has only the virgin islands.
Mexico? Nothing at all. You have none of these milestones ready so you're decades away from ever getting your valley.
Step 1. Port Vallarta becomes a tech free trade zone.
Step 2. A major electrical investment that makes electricity in that region essentially free to everyone and highly reliable.
Step 3. Tooling for automated tech is subsidized, but restricted from leaving the region.
LLMs are more than left vs right brained. They have multiple experts; they have multiple modality.
Some big models can have hundreds of "side of brain"
print("excellent choice.")
else:
print("need to fix language")
I'm far more willing to try new libraries or new frameworks.
>Personally, I've found that the paradigm of "sharp knives in the drawer" can lead to some pretty nasty output from LLMs; it's felt to me like the higher the ambiguity the higher the variance in quality.
I learnt this one when trying to do video game programming. The models just arent as competent at more niche programming.
Something to also add to your list. Context management seems to make me want to break out to utility files. Keep your files small and the llm will agentically only be looking at smaller files. Using up less context overall. Rarely does an LLM need to have the entire project in context, but better yet the smaller you keep files, the more reliable it becomes. As accuracy is lost the higher the context because it tends to forget about context.
I'm the AI expert for my org. Everyone else is more or less opposed to AI.
>One senior-dev (team-lead also) tried to explain to me that AI is a subfield of machine-learning, and always stochastic in nature (since ChatGPT responds differently to the same prompt).
machine learning is the sub field of AI.
Not really stochastic as far as I know. The whole random seed and temperature thing is a bit of a grey area for my full understanding. Let alone the topk, top p, etc. I often just accept what's recommended from the model folks.
>We/they are selling tailor-made "AI-products" to other businesses, but apparently we don't know how sampling works...?
Sales people dont tend to know jack. That doesnt mean they dont have an introvert in the back who does know what's going on.
>Am I just too junior/naive to get this or am I cooked?
AI for the most part has been out a couple years. With rapid improvement and changes that make 2023 knowledge obsolete. 100% of us are juniors in AI.
You're disillusioned because the "ai experts" basically dont exist.
Right thread, but I was kind of placing the context of my needs and what hasnt worked. Just not good enough.
>I would normally be wary of using any remote AI solutions since some of the documents that I am thinking of storing contain very personal information but, llamaindex seems to support local-llms. https://developers.llamaindex.ai/python/framework/getting_st...
llama index can be fully local. In fact, i believe the size of the llms are trivial and dont need to be large.
Langchain does have a ruby port if you want, but even if you run the python version which is superior it doesnt impact using it as a service.
Personally I'd go more llamaindex route.
There are newer options, opendevin, autogen, crewai, smolagents. I dont know much about them tbh.
My backup if that hadnt worked. You get a mosquitto outdoor zapper light. You put it on a timer so that it runs ~3-4 hours every night, mainly this is during twilight time. So 2 hours just after sun down, 2 hours before sun up. Every night that it runs, at the end of the season you'll be like 90% less mosquitos.
The reality is that none of them are particularly good. Obsidian is my current solution, but i dont like it.
What I really need to give a chance: Llama Index. https://developers.llamaindex.ai/python/framework/#use-cases
Obviously I'd need some time to customize it for my needs, but it sure does seem to be what i want.