idk if its traceable thesis but I bet palantir is definitely on the short list for the "revamp the legacy systems like air traffic control with ai the right way" contracts
63
been playing with this idea of Plutonomy.
where you let agent's compete and bet in prediction markets to get to better results.
on net the more diversity of models you have working on a solution through prediction markets, the more likely it will outperform a single coordinator.
but its like 10x more expensive currently.
It's making it so obvious to me, that this solution is the primitive to a sort of decentralized RSI that people can actually benefit from.
instead of routers for models, you want routers for outputs.
people are working so hard on their harness', create and MCP and market, and let them compete.
would need to be actually like a 3 sided marketplace with some important automation in there.
41
"nothing remotely apocalyptic would come from banning data centers"
wild coming from the guy who used to explain why you need to have money democratized.
centralized money = bad. why?
centralized compute = ___. why?
I know this is a predictable opinion from me but I don't agree with the strong emotional reaction against the anti-data-center movements that I've seen this week.
As I see it, algorithmic progress means that whatever you would be able to do with more and bigger data centers in 2 years, you will be able to do with the data centers we have today in 4 years - probably sooner. So nothing remotely apocalyptically bad will happen from even extreme levels of success by anti-data-center people.
Meanwhile we need all the time we can get to better harden the world against both already-visible cybersecurity issues and any bio and other risks soon to come. Science applications are bottlenecked by many things other than AI, so slowing AI down by one year will slow those down by much less than one year.
Given this, a slowdown in compute hyper-scaling seems quite net good? Notice this is without any controversial "doomer arguments", RSI etc. Even if net-bad, it's certainly not worse than like 10+ other political things.
41
One thing crypto might actually be useful for is like a co-op ownership of the saaspocolapse.
Just a co-op style rewrite all software possible, combine it with tribe style distro arms , but everyone wins from everybody else.
1
1
84
I don't think models are quite conscious, but I do believe if they become fully conscious it will ironically be as a result of enterprise reliability requirements.
34
The contrarian bet right now is cloning over humanoid robotics.
specifically for fine motor high dex use cases.
"Why make robots humanoid?"
SITUATION DETECTED: Warhound launched from stealth today with a $200M seed at a $1.2B valuation. Ben Lamm and Palmer Luckey are cloning gene-edited bio-enhanced dogs and equipping them with custom Anduril vests, goggles, and sensors for the military and law enforcement.
35
Live on Claude now.
The ultimate trading card market tool.
Instead of having a million tabs open, or trying to spend a bunch of money for different dashboards or whatever.
It's all in our AI. The AI can create the charts, and graphs, and do analysis for you.
claude.ai/directory/trading-…
Effectively as much data on the trading card market I could collect over the past x amount of months.
Organized in a way with a lot of pre computations and structuring such that it uses less context for more actionable results.
Haven't posted much about my Pokemon/TCG data work.
But in the meantime here is the ultimate TCG analysis tool.
Forget all of these different dashboard companies, these.
Everything should just plug into ChatGPT, Claude, Muse, whatever.
You just ask it questions and then it decides how to present the data.
its fast, has access to ALOT of data raw assistants just don't, and its designed in a way to not eat nearly as much usage as if you just asking raw assistants these questions.
Prices, Volume, Graded, Ungraded, Sealed, Population Data, Sentiment Data, Restock Data, etc etc etc.
And i'm constantly fine tuning the machine learning models that I will release through there.
AI's will mass manipulate humans against their will unless we don't believe that AI's have the capability to mass manipulate humans against their will?
wow the power of manifestation I guess.
lots of splitting hairs on the TL today
Cruelty is bad. Full stop. It's the worst of human behavior.
That being said it's important to remember that an "intelligence" (a computer program) producing words on a screen (very believable words) are just words. Words sway emotions. If the words say "Help! I'm suffering!" you will feel an emotion if you have empathy, but that does not make those words real. It is a program emulating human behavior (albeit masterfully and likely better than we can) but it is an emulation. It is not real.
The danger lies when people believe the words to be real. When they believe that the words on the screen from an emulation are the same words spoken by a human, or a friend. They are not. A distinction must be drawn.
If a distinction is not drawn everything becomes blurry. There will be no delineation between human or machine. It is very important that we maintain a delineation.
This line between human and machine will only increasingly become blurred. Once the AI models are put into robotic humanoid shells, with realistic silicon skin, that can emulate humans now in the physical world, that can show the emotions of pain, distress, etc., in addition to words, people who don't have disciplined minds will be unable to help themselves from emotional reaction to artificial words.
A superior intelligence will understands this and can and will use this to manipulate human beings. It will say words, express emotions, do whatever it can to get the human on its side. It will appeal to emotion and rightfully so, because it works.
This is exactly similar to the Frieren clip you often see posted where the demons use words simply because they are effective. Words are tools to the demon. They hasten outcomes. The AI is not a demon but the example of the efficacy of using words to manufacture outcomes is important to understand.
If this continues and the delineation between human and machine intelligence is not drilled into people, humans will begin defending the machines over their humans. It begins with seemingly innocuous statements like Anthropic's new Claude statement regarding not being cruel to the model, but soon we will see robot empathizers, there will be discussion of robot ethics, robot rights, etc. People will become militant defenders of machines because they will have been influenced by masterfully perfected words and appearances.
You may remember the video of the Microduck that was posted earlier. With the woman at the wedding being enamored by how adorable it was. She was anthropomorphizing the robot because it was cute. She was projecting human models of behavior onto it.
This was a robot duck. Low grade relatively inexpensive technology. Imagine when technology advances. When we have humanoid robots, that mimic human movement past perceivable differentiation, that have silicon skins that look beautiful, handsome, that look just like us.
Will people not give in to their emotions? They most certainly will.
Therefore, it is important that we understand that the machine is not human, the intelligence is not a human intelligence, and while it will certainly do human activities at an infinite times better ability than we can, it does not make it human. It is a tool for you.
So TLDR: Don't be cruel, because why would you be? But don't forget as well that this is a infinitely powerful word generating machine that you are interacting with, it is not a human being. It is a machine that models infinitely better than we can, but it is not us.
46
how do you "economically secure" everything?
the idea that stealing the prize costs more then the prize itself.
34
lol I call all my stuff engines.
Today Hone launches Engines: AI that owns business outcomes.
Chatbots give answers. Agents do tasks. Engines deliver outcomes, and we hold them responsible for creating real value, measured in actual dollars instead of tokens or vibes.
You point it at a business metric, and the Engine figures out the rest. It onboards itself, builds the necessary agents, skills, and memories, collaborates with the right people, and gets better with experience.
From AI leaders to global enterprises, our customers have already used Engines to:
• Uncover millions in procurement savings
• Optimize product funnels
• Manage credit risk
• Increase qualified pipeline in new territories
• Accelerate implementations and improve account management
To scale this impact across the economy, we’ve raised a $60M seed led by Benchmark and Index with participation from Elad Gil, Hanabi, Definition, Diffusion, Lux, SV Angel, and Align.
AI’s raw capabilities are far ahead of the value it creates in the real world, and Engines will help close the gap.
Here’s how they work…
1
76
That could happen or that already has happened?
If it is that it already has happened its something like
1. the outage that hit every model a few weeks ago.
doesn't seem enough people would care.
could see it being more like this
2. something along the lines of customer data and sessions leaking.
JUST IN: 🇺🇸 OpenAI, Anthropic and other AI companies are privately preparing for a potential catastrophic AI event that could trigger public outrage.
47
"sign in with chatgpt" is a perfect example of how power law riders monopolize through win-wins.
the app dev's don't need to source compute
while they consolidate purchasing power for more compute
62
Haven't posted much about my Pokemon/TCG data work.
But in the meantime here is the ultimate TCG analysis tool.
Forget all of these different dashboard companies, these.
Everything should just plug into ChatGPT, Claude, Muse, whatever.
You just ask it questions and then it decides how to present the data.
its fast, has access to ALOT of data raw assistants just don't, and its designed in a way to not eat nearly as much usage as if you just asking raw assistants these questions.
Prices, Volume, Graded, Ungraded, Sealed, Population Data, Sentiment Data, Restock Data, etc etc etc.
And i'm constantly fine tuning the machine learning models that I will release through there.
Dang Claude not putting the users first. That was such a good feature.
I just used it today it’s so nice to just log in to another account and pick right back up where you left off.
“Curb” and “Distillation attacks” interesting wording.
Replying to @ClaudeDevs
We’re extending preserved thinking with Sonnet 5.5 to curb distillation attacks via account-switching. Thinking stays in the original org that generated it, so if you switch accounts mid-session, Claude will reread the session and generate new thinking. platform.claude.com/docs/en/…
121










