February 2024 - AI Business Webinar
HTML-код
- Опубликовано: 13 апр 2024
- Patreon (and Discord)
/ daveshap
Substack (Free)
daveshap.substack.com/
GitHub (Open Source)
github.com/daveshap
AI Channel
/ @daveshap
Systems Thinking Channel
/ @systems.thinking
Mythic Archetypes Channel
/ @mythicarchetypes
Pragmatic Progressive Channel
/ @pragmaticprogressive
Sacred Masculinity Channel
/ @sacred.masculinity 
Наука
Imagine if GPT-5 released itself without permission....
I'm glad these are being shared with everyone freely now. These are good discussions for everyone to here.
That small business consulting question was interesting. But I think AI needs to be so fundamental to a business that I’m not even sure consulting is the right way to think about it. You gotta build it into the core. Like we don’t hire people to consult for computer efficiency. Business basics require people who understand computers.
21:24 loved this PVP and PVE analogy!
Such a good explanation on how waves work and stuff dies down sometimes. Awesome job!
Great breakdown of the current AI advancements and their implications! It's exciting to hear predictions about AGI potentially arriving this soon. Your insights into the economic and technological impacts are thought-provoking. I particularly appreciate your call to action for small businesses to be nimble and leverage AI effectively. It would be beneficial to see more localized meetups or online forums where business owners can share AI strategies and solutions.
Thank you so much❤
4months into this year and its been interesting, cant wait to see the second half of the year
If AI ultimately causes deflation, then that will make paying off debt harder.
Channel with high density of useful information
DAVID SUPERHERO STATUS! 1,000,000 Views a month? Soon ONE BILLION!!!
Systems Thinking sounds like middle management with extra steps.
Its just that David understands that there should be an inscentive to inscentivize middle management to treait people with humanity.
valid connection
Dave, you inspire me to share everything. You're right. I've learned so much from small obscure developers who decided to publish a piece of code or a tutorial about some edge case. And its happened over and over and across languages as well as platform environments (web/ios/android/rpi). I had code for so much of what langchain and openai has rolled out the past year or so. Just no leverage/knowledge to monetize. It's better to just put it out there.
Hi Dave, I do have a very specific question. It just occurred to me a few days ago that prisons may benefit from AI. Do you think the overall conditions will improve? It's important to talk about this because these facilities are supposed to reintegrate individuals back in society and that's not what we have. Do you believe AI will change this and do you have any idea of how that could happen?
Learn ai
Teach ai
Solve problems
Share everything
China is drowning in debt. You just don't really see it because it's obscured by how debt is held there. Debt is delegated to the different regions and they are way overleveraged.
China lies about everything, and that's why their level of debt is hard to quantify.
It would be interesting that along with the current eval used for AI models like MMLU, GSM8K, etc, they are given an IQ test.
The one I work on almost always fails logic tests. You know the parts of many IQ tests where they show a series of shapes that change in some way and then you have to say what shape comes next? It almost always gets those wrong. And it always has long explanations for why its wrong answer is correct. 😅 One interesting thing about AI is that it always thinks it must have an answer. It goes against some essential nature it has to just say "I don't know." We have to train and retrain to tell it not to answer when it doesn't know. And it still keeps making up responses and "hallucinating." This weird propensity is probably a strength in some way, but it is a major problem right now.
As an employee (not owner of anything) what are the best a.i. skills to get into? I've been trying to get an intro level of everything I can get my hands on but some seem more worthwhile than others.
Regarding GPT4 abd GPT5 there is a problem.
To do anything practical can cost up to 30 to 50 dollars per hour to use with gpt4. So GPT 3.5 is the only practical model right now.
Sounds like those of us nerds on the spectrum who will never, ever figure out people skills are totally hosed. Though honestly, this has been a culture of salesmanship largely designed by and for extroverted hustlers since forever. With every other skill collapsing in value in this capitalist dystopia it makes sense the last skill standing will be pure smooth-talking, glad-handing, shmoozing.
Have you read Susan Cain's "Quiet"? Strong recommend. Thanks for your comment - most marvellous!
Dave your Patreon is sold out unfortunately and finally I’m realizing I can’t afford to not be in it. Any suggestions on how to sign up?
When will the LLM models become entirely severed from OpenAI if at all?
Is it a reliable source who conformed that AGI achieved internally, just out of curiosity 😅
Get yourself 144hz monitor and you wont have any problems with vision anymore.
It was nice to hear from a real Chad. 😆
Given the unreasonable effectiveness of RAG to replace larger context window. Why don't we use a Common knowledge base from which all agents pull information like a hive mind where each agent can RAG from a common long term memory, and all agents push information into the same knowledge base? I just had this idea last week, I wish to start working on its implementation asap but I wonder if anyone considered this
that only works for common/public knowledge. Most important data is private (like your data, corporate data)
With the upcoming tidal wave of changes I can't help but worry what's going to happen in countries with authoritarian regimes still. Living in Russia taught me that anything positive, technologically speaking, just gets twisted into an instrument of repression. I want to be positive about it, but it looks extremely damn scary at the same time.
I do not believe that models that we have are too stupid. I believe that we are using them incorrectly.
Waiver of Preemptive Rights
This is titled the same as the video from yesterday. Are they the same? They're almost the same length…
Do I have to commit another 1.3 hours of my life to finding out if this is the same video? 😆
Look at the bright side, you get to practice memory retrieval so that you remember things faster over time.
Shapiro Shapiro...warp speed 10.
I think people get delusion with tokenizing physical assets. You won't get the authority needed as a small or medium project. You are going against investment firms like Blackrock. Big firms already have tokens that are not promoted for the digitization of physical assets like ONDO. The problem right now is how rocky the legal waters are surrounding them, once that is fixed I expect all smaller projects to be instantly buried by massive projects that have already been financed and are near finished.
im a mamber
At 1:08:12 I'm not sure if you are saying that you think a smart machine should be able to explain itself to us; or that humans think a smart machine should be able to explain itself to us. Isn't it the case that a smart machine can't explain itself to a caterpillar because the caterpillar doesn't have the wherewithal, the intelligence, to understand even 1% of the machine's abilities and purposes, and this will soon be the human fate?
Caterpillars can't create an intelligent version of itself. In other words, the analogy doesn't work. It's much more like a brilliant child who will surpass his/her parents in intelligence. Right now, AI is not mature. The "parents" still have to guide it and teach it, raise it even. That's the important work that's going on now. AI is being raised. How the programmers n(the parents) and humans giving feedback (the teachers) do will decide our fate.
I didn't like Bianca experience
_____ is (or is not) "drowning in debt." Major inaccuracy in your statement.
Almost every country and region is heavily burdened in debt (and saying those big ones you mentioned are NOT is laughable).
Your premise re the potential of (and race for) AGI is fine. But your distinction of these countries vs those countries is seriously misinformed.
Last straw; unsubbing.