Ari Paul@AriDavidPaul · 6 天前赛道影响Fairly soon we’ll likely be recording (almost) all conversations by default. Listening devices everywhere makes it happen passively, and then we’ll want the benefits of automated note taking and follow up by our AI assistants.
How does this change life and society? I think it will be pretty dramatic. Consider arguments (both personal and professional) in the context of an instantly searchable history of every prior convo.
Related - we’ll likely get an extensive bio of someone upon first meeting them (whether via smart glasses or something else.) And this bio isn’t just a resume, it could include their spiciest social media posts if you wanted. How does that change social gatherings? You bump into someone at a rave or sporting event and can see…anything relevant from their public history.
AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。
Ari Paul@AriDavidPaul · 6 天前观点输出The mayor of Nice describing the reality below. We’re being told these are “students” protesting government funding…the attackers say otherwise.
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Ari Paul@AriDavidPaul · 8 天前观点输出My first job out of college was as an options market maker at SIG. I have a CFA, MBA, and I traded the first ever Bitcoin VAR swap. I’ve been a net loser in equity options over my lifetime, and that was without any blowups or big mistakes. Better to just avoid them (rare exceptions.) Buy good assets and hold them until A. You find much better assets or B. Those original assets are clearly overvalued.
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Ari Paul@AriDavidPaul · 12 天前观点输出It’s very very hard to be “private.” Shielded crypto txs just one small part of the puzzle. Moving money anonymously, is imo, beyond the capability of most of us today. For “small fish” we used to have privacy through obscurity. But automated tooling potentially eliminates that and can de-anonymize most txs.
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Ari Paul@AriDavidPaul · 21 天前赛道影响Following up on this. Turns out I have light AuDHD, which was incredibly helpful to learn. That came after testing clearly negative for everything and seeing multiple psychiatrists. That diagnosis came from a mix of my own research + Claude and has since been confirmed by a neuroscientists and psychiatrist. It gave me a new lens on challenges I was facing. For example, it reframed “depression” as mostly “autistic burnout” for me, which requires a different remedy; (activity for depression, genuine rest for burnout). That diagnosis also led me to other things, like a diagnosis of light hypermobility. Neurodivergence is still poorly understood, but it’s foundationally connected with overactive immune system (specifically MAST cells, histamine sensitivity), and stomach issues in many. For myself, I just have light seasonal allergies, a slight histamine sensitivity, and light hypermobility (double jointed elbows, hypermobile spine in spots.) Main take away from this journey is to keep digging and use LLMs for self-diagnosis (and then verify with the right experts.)
The diagnosis was helpful to me, because I’ve long experienced the inherent contradictions and “back and forth” between the adhd and autism pulls. I didn’t know how to describe or model it, and so just thought of myself as someone with inconsistent energy and a weakness for procrastination at times. The diagnosis let me reframe that as different nervous system states, and quickly yielded specific “hacks.” For example, if I’m procrastinating on something now, I’ll often “speed task” it, or in some other way gamify it to “trick” my adhd brain into tackling it immediately.
It also helped me understand why some activities were surprisingly draining for me. “Masking” is exhausting.
I think I’m 90% recovered from burnout after about 21 months into “retirement.” That was a combination of:
“Touching grass” - lots of time in nature and ideally in flow state via hiking or climbing or meditating.
Connecting with friends and family - spent a lot of time with family and my ~10 closest friends, quality time on what matters. Love and friendship. Looked for ways to help others.
Transcranial magnetic stimulation: did 2 rounds of this, each time it cut my anxiety by 35% (sustained), and cut 5 years of brain fog each time.
Supplements and peptides: taking “Wolverine stack” and a bunch of supplements now, but noticed very little from any of it. With wolverine stack i recover from workouts a little faster. No affect on mood or burnout from any of it. But my blood test results were solid at the start; worth getting a blood test, if you’re low on something, could be an easy fix.
42
Ari Paul@AriDavidPaul · 22 天前观点输出Climbing rocks and trees in Quebec. I miss fall colors in Miami. https://t.co/bCFliatStA
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Ari Paul@AriDavidPaul · 23 天前政策影响Nearing the security rubicon. The world isn’t ready to deal with this.
AI：偏政策推进，强调合规落地与规则明晰，对监管主线更敏感。
Ari Paul@AriDavidPaul · 24 天前观点输出Hi from Quebec. https://t.co/9sw7lGzRZN
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Ari Paul@AriDavidPaul · 27 天前政策影响The legal systems of the world (all of them) were built for a very different era. They’ve long been failing at “diffuse harms at a distance”, like assessing accountability for pollution that probabilistically caused cancer deaths 1000 miles away.
Now I’m seeing a lot of chatter around the legal system and OpenAI and Anthropic’s possibly criminal hacks of companies and governments. Some are arguing existing laws work fine and just need to be enforced. I don’t think so. Laws related to hacking were already vague and capriciously enforced. One big problem is intent - any law or regulation that mentions intent gets murky with AI. We’ve already seen this with trading algos - they’re functionally able to front run and manipulate markets with legal impunity, since “market manipulation” and most “trading crimes” requires intent, and algos and AI don’t have intent (or at least that’s a relatively new and untested question.)
So we’re currently in a world where there’s a lot of stuff that’s criminal for you to do…but legal for you to ask an algo or LLM to do it, as long as your instruction is vague enough and a prosecutor can’t prove you knew ahead of time exactly what would result from your request.
Another angle is the failure of ex poste enforcement when harms get big enough. Punishing suicide bombers is obviously useless. Similarly, it doesn’t do the world any good to punish someone after they nuke and kill and 100m people. Those harms are too great to just punish after the fact. And there are too many people alive who will happily risk life in prison or the death penalty to kill 100m of their enemies, or to steal $100b.
An AI hack can potentially steal from or majorly harm 1 billion people instantly. Relying on a court to maybe remediate that years later doesn’t works.
IMO, this will be a huge multi decade undertaking to create new legal primitives suitable for the modern world. It’ll be a painful transition.
Ari Paul@AriDavidPaul · 29 天前赛道影响With all the concerns of AI doom, we can’t forget about the real risks facing America. Furries. https://t.co/zyiRMavC8X
AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。
评论
暂无评论