Alex Epstein
speaker
366 appearances
2 recordings
2 series
first heard Jan 2025
last heard 28 Apr
Alex Epstein’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsRecordings per month over the last 12 months — 1 in all, peaking in Apr 2026 with 1.
Appearances
So it's just this thing where, but I noticed these trade groups, like I was in a meeting recently and another person in the meeting might be way more famous and way richer than I am, but I feel like everyone in the room trusts me more because they know I'm saying what I think is right, which doesn't mean they'll agree with me.
Yeah. In this case, I am right.
There is also that. No, no. But yeah, I know. It's like and just knowing that you can if the trade groups really came across like we really believe this and we're saying this not just because we're in oil or solar or whatever. We're in oil or solar because we think it's good and we're going to only tell you what the truth is.
Even when it causes us to lose some short term subsidy.
They would be so disappointed. powerful yes amen to that that's definitely the case yeah so i forget oh yeah i was just saying you're the chat gbt yeah like it came up with this and i just thought oh wow this is going to be really fun in the messaging because you can just it's uh and even some of the canadian stuff i've been doing i've been using it so yeah
People need, if you're in any kind of intellectual thing or really anything that relies on high quality decisions, you need, here's my free advertisement, Sam Altman, like you should use something like chat. You should just get whatever the cutting edge thing is because it's so much cheaper than people. And it really is replicating and replacing a bunch of human functionality.
Is buying intelligence really cheaply a good thing? Right. Right, right, exactly.
Yeah, and people are just so... Like, it's so... This is one final digression. People are just so... They just don't yet realize how efficient it's going to be to use these things. Not every application, by any means, but for many applications, they have all these things like, oh, it doesn't always get directions proper. Oh, really? All your human employees always get the directions correct?
Right. Like at this point, ChatGPG Pro is better than almost any human you will ever employ in terms of following directions. Like it is really, really good in terms of just if you write out what you think, if you say it in a circuitous way, it is pretty damn good. And in terms of like the output, this is another thing I think people don't get is they're like,
For instance, I'll give you an example. I was running this CO2 thing, like this natural gas CO2 thing, and I ran it with the non-pro version of ChatGPT, and it gave me a false thing that was something like, I wanted to know how does the volume of natural gas compare to the volume of CO2 generated, right?
Because one way to think about this idea of let's use fossil fuels and capture the CO2 is you need to build a whole new industry for the captured CO2. So if it's a one-to-one ratio, right, if it's a one-to-one ratio of natural gas and CO2, then you need to basically double the size of the industry, right, in terms of piping it and putting it underground. Right, right.
And I thought it was a one-to-one, and then I was in the airport and talking to ChatGPT Plus, and it told me it was like 59 times. I'm like, that sounds wrong, but if that's true, that's crazy. And I was sort of excited, like, oh, wow, that's a blockbuster. But then I was like, I ran it the other day with ChatGPT Pro, and it said, no, no, it's one-to-one, which made more sense to me.
But it's like, okay, yes, sometimes it'll make a mistake like that, but I don't go to print with that thing without confirming it by an expert. Yes, yes. But even if it's right 95% of the time or 98% of the time, you can often get, I don't have the perfect words for this, but you can often get like the shape of what the solution will look like, even if not every value is correct.
So you can get the idea of, yeah, you do need to build a whole new parallel industry.
cost a lot of money even if it's your estimate is 50 times you can help think through it so just the fact that it can make errors a humans can make errors too but it can really help you explore territory and develop ideas very very quickly and then what you might find is that you make one or two errors at the end but usually those are not decisive errors that throw out the whole project usually because every kind of thought process involves lots and lots of little
uh assumptions and and interrelationships and if you get 98 of that right probably the overall thing is going to be right yeah probably a very small percentage of the time it's going to be oh there was one false assumption and the whole thing is wrong i have yet to find It could happen. It could happen. So, yeah, I think people are just underestimating how good this stuff is going to be.
And they may be even overestimating certain things. And it could be we're back in five years. And, yeah, there might even be a bubble in certain ways. But at this point, it's obvious that for a lot of people, it's going to help them. And I would just say, as an entrepreneur who runs a small business that's very intelligence-based, obviously helps me think much, much better.
Oh, yeah, I should mention alexepstein.ai is now free, so people should check that out. Oh, yeah, okay. So that's now the latest version of Energy Talking Points. We're already having some elected officials use this for, like, floor speeches. You can have Alex AI write you a floor speech. And so we've spent a lot of money customizing an AI.
Yeah, .ai.
And what it's really done is it's engineered with prompting that is very based on this even-handed and pro-human approach Right, so you've built that ethos into it. Yes, and really one thing it does really well is question assumptions. So it scans everything for, does this question or statement have an assumption that Alex Epstein would disagree with?
Showing 201–220 of 366 · page 11 of 19
← Previous
Next →