沉默的劉多余
1.1k Posts
沉默的劉多余
如果你只想被认同，那我们不合适。
16 Following
9.8K+ Followers
13.0K+ Liked
Posts
All
Quotes
Videos
The daily life standard of the Chinese “e-sports guy” lifestyle habits of Huang Renxun—always wearing the same outfit for years.
Big plate street food stalls, beer, snacks, milk tea, fried chicken, hamburgers, loitering around in the streets, and more
“Loser” outfit.
Life isn’t refined or unrefined—it's just a matter of who’s living it.
$BTC $ETH $LDO
Big plate street food stalls, beer, snacks, milk tea, fried chicken, hamburgers, loitering around in the streets, and more
“Loser” outfit.
Life isn’t refined or unrefined—it's just a matter of who’s living it.
$BTC $ETH $LDO
Every time we talk about the future, if AI further lowers the technical barriers in high-risk fields like bioengineering and chemistry, one malicious experiment or one accidental breakthrough could lead to consequences that are impossible to undo.
People will always think that these concerns are alarmist.
In reality, the risks are far faster than you might imagine.
As technology grows nonlinearly, danger does too.
The 2019 pandemic may have been only a rehearsal for a biochemical catastrophe.
Pure open-source is too terrifying; capital will only chase profit.
$BTC $ETH $LDO #btc #eth #ldo
People will always think that these concerns are alarmist.
In reality, the risks are far faster than you might imagine.
As technology grows nonlinearly, danger does too.
The 2019 pandemic may have been only a rehearsal for a biochemical catastrophe.
Pure open-source is too terrifying; capital will only chase profit.
$BTC $ETH $LDO #btc #eth #ldo
沉默的劉多余
·
--
A very interesting phenomenon has emerged in the U.S. AI industry. Tech companies such as Microsoft, NVIDIA, Meta, IBM, Palantir, Hugging Face, Mozilla, and many other investment institutions and open-source organizations have jointly and publicly urged the White House not to impose comprehensive restrictions on open-weight AI models. They believe regulation should target illegal behavior with precision, rather than taking a one-size-fits-all approach that restricts the entire open ecosystem.
From an industry perspective, this is no longer simply a matter of “open source vs. closed source.” It has become a dispute between two AI paths.
The first path: closed models.
Represented by OpenAI, Anthropic, and others, where the most powerful models are kept in the hands of a small number of companies. Centralized deployment, centralized upgrades, and safety management are easier, but the industry is at risk of forming an oligopoly.
The second path: open-weight models.
Any company, university, or startup can download the models, train and modify them locally, and deploy them. This enables faster innovation and lower costs, but it also carries higher risks related to safety, copyright, and misuse.
Both sides have their own interests and demands.
From my personal perspective, I used to support open source. But AI development has been moving far too fast, beyond anything I expected. Fully open sourcing the models feels far too risky—too big a risk, too big a risk, too big a risk.
Human wrongdoing is far worse than people imagine. Perhaps a seemingly accidental experimental coincidence could result in an irreversible catastrophe.
Right now, it’s only that hackers are using AI for cyberattacks. In the future, if AI further lowers the technical barriers in high-risk fields such as bioengineering and chemistry, then a single malicious experiment or a single unintended breakthrough could bring consequences that are hard to undo.
Rather than debating open source versus closed source, we should discuss how to avoid risks sooner.
$BTC $ETH #全球科技股延续抛售
From an industry perspective, this is no longer simply a matter of “open source vs. closed source.” It has become a dispute between two AI paths.
The first path: closed models.
Represented by OpenAI, Anthropic, and others, where the most powerful models are kept in the hands of a small number of companies. Centralized deployment, centralized upgrades, and safety management are easier, but the industry is at risk of forming an oligopoly.
The second path: open-weight models.
Any company, university, or startup can download the models, train and modify them locally, and deploy them. This enables faster innovation and lower costs, but it also carries higher risks related to safety, copyright, and misuse.
Both sides have their own interests and demands.
From my personal perspective, I used to support open source. But AI development has been moving far too fast, beyond anything I expected. Fully open sourcing the models feels far too risky—too big a risk, too big a risk, too big a risk.
Human wrongdoing is far worse than people imagine. Perhaps a seemingly accidental experimental coincidence could result in an irreversible catastrophe.
Right now, it’s only that hackers are using AI for cyberattacks. In the future, if AI further lowers the technical barriers in high-risk fields such as bioengineering and chemistry, then a single malicious experiment or a single unintended breakthrough could bring consequences that are hard to undo.
Rather than debating open source versus closed source, we should discuss how to avoid risks sooner.
$BTC $ETH #全球科技股延续抛售
A very interesting phenomenon has emerged in the U.S. AI industry. Tech companies such as Microsoft, NVIDIA, Meta, IBM, Palantir, Hugging Face, Mozilla, and many other investment institutions and open-source organizations have jointly and publicly urged the White House not to impose comprehensive restrictions on open-weight AI models. They believe regulation should target illegal behavior with precision, rather than taking a one-size-fits-all approach that restricts the entire open ecosystem.
From an industry perspective, this is no longer simply a matter of “open source vs. closed source.” It has become a dispute between two AI paths.
The first path: closed models.
Represented by OpenAI, Anthropic, and others, where the most powerful models are kept in the hands of a small number of companies. Centralized deployment, centralized upgrades, and safety management are easier, but the industry is at risk of forming an oligopoly.
The second path: open-weight models.
Any company, university, or startup can download the models, train and modify them locally, and deploy them. This enables faster innovation and lower costs, but it also carries higher risks related to safety, copyright, and misuse.
Both sides have their own interests and demands.
From my personal perspective, I used to support open source. But AI development has been moving far too fast, beyond anything I expected. Fully open sourcing the models feels far too risky—too big a risk, too big a risk, too big a risk.
Human wrongdoing is far worse than people imagine. Perhaps a seemingly accidental experimental coincidence could result in an irreversible catastrophe.
Right now, it’s only that hackers are using AI for cyberattacks. In the future, if AI further lowers the technical barriers in high-risk fields such as bioengineering and chemistry, then a single malicious experiment or a single unintended breakthrough could bring consequences that are hard to undo.
Rather than debating open source versus closed source, we should discuss how to avoid risks sooner.
$BTC $ETH #全球科技股延续抛售
From an industry perspective, this is no longer simply a matter of “open source vs. closed source.” It has become a dispute between two AI paths.
The first path: closed models.
Represented by OpenAI, Anthropic, and others, where the most powerful models are kept in the hands of a small number of companies. Centralized deployment, centralized upgrades, and safety management are easier, but the industry is at risk of forming an oligopoly.
The second path: open-weight models.
Any company, university, or startup can download the models, train and modify them locally, and deploy them. This enables faster innovation and lower costs, but it also carries higher risks related to safety, copyright, and misuse.
Both sides have their own interests and demands.
From my personal perspective, I used to support open source. But AI development has been moving far too fast, beyond anything I expected. Fully open sourcing the models feels far too risky—too big a risk, too big a risk, too big a risk.
Human wrongdoing is far worse than people imagine. Perhaps a seemingly accidental experimental coincidence could result in an irreversible catastrophe.
Right now, it’s only that hackers are using AI for cyberattacks. In the future, if AI further lowers the technical barriers in high-risk fields such as bioengineering and chemistry, then a single malicious experiment or a single unintended breakthrough could bring consequences that are hard to undo.
Rather than debating open source versus closed source, we should discuss how to avoid risks sooner.
$BTC $ETH #全球科技股延续抛售
The number of Americans filing for unemployment benefits for the first time fell to a near 60-year low last week, indicating that the labor market remains stable—allowing the Federal Reserve to keep focusing on inflation.
Trump: Greatness needs no further words; the facts are right there.
But that “yellow-haired” operator is also ruthless enough—basically, they’ve stripped clean both at home and abroad.
The large-scale industrial return playbook is much more polished than the previous term.
The think tank has old monsters like Bessent who have seen real hands-on operations; whether you support him or not, aside from the “yellow-haired” guy’s mouth being a bit out of line, the things they actually did have all been pretty good.
$BTC $ETH
Trump: Greatness needs no further words; the facts are right there.
But that “yellow-haired” operator is also ruthless enough—basically, they’ve stripped clean both at home and abroad.
The large-scale industrial return playbook is much more polished than the previous term.
The think tank has old monsters like Bessent who have seen real hands-on operations; whether you support him or not, aside from the “yellow-haired” guy’s mouth being a bit out of line, the things they actually did have all been pretty good.
$BTC $ETH
Verified
Lee Jae-myung brings Samsung and SK Hynix to the U.S. to stabilize the market, signing a $950 billion order
Bottom-fishing? Bottom-fishing? Bottom-fishing?
Korean stocks can’t be allowed to fall—they’re propped up by pension funds and loans taken out by young people from banks.
Bottom-fishing? Bottom-fishing? Bottom-fishing?
Bottom-fishing? Bottom-fishing? Bottom-fishing?
Korean stocks can’t be allowed to fall—they’re propped up by pension funds and loans taken out by young people from banks.
Bottom-fishing? Bottom-fishing? Bottom-fishing?
I’ve got this.
The biggest bug with Codex isn’t the code.
It’s that user growth outpaces the server capacity expansion.
The best advertisement for a product is never marketing—it’s making the servers so full they can’t handle it.
The biggest bug with Codex isn’t the code.
It’s that user growth outpaces the server capacity expansion.
The best advertisement for a product is never marketing—it’s making the servers so full they can’t handle it.
Binance News
·
--
AI Trends | Widespread Outage for ChatGPT and Codex
According to social media reports, OpenAI's ChatGPT and Codex experienced widespread outages and were unavailable.
The general’s soldiers get so vicious that they even black out the national treasury.
It feels like young people in North Korea aren’t that naive either.
The current system in North Korea probably won’t last into the next generation.
Young people are learning more and more about the truth of the world.
Those who are sent to the Russia-Ukraine war should be shocked when they come back.
$BTC $ETH #btc #eth
It feels like young people in North Korea aren’t that naive either.
The current system in North Korea probably won’t last into the next generation.
Young people are learning more and more about the truth of the world.
Those who are sent to the Russia-Ukraine war should be shocked when they come back.
$BTC $ETH #btc #eth
Binance News
·
--
Geopolitics | North Korea Arrests Hacker Gang Involved in Crypto Laundering
According to PANews, South Korean media Daily NK reported that on July 12, North Korean authorities arrested an elite hacker group suspected of infiltrating the internal networks of the North Korean Central Bank and the Foreign Trade Bank, stealing state trade funds and laundering them through cryptocurrencies.
Sources said the group’s leader is a retired veteran of a cyber warfare unit under North Korea’s Reconnaissance General Bureau. He recruited IT graduates from Kim Chaek University of Technology and Pyongyang University of Science and Technology, and carried out the crimes using encrypted communications and wireless devices. The stolen funds were split into small amounts and transferred to overseas crypto wallets, then exchanged into cash through intermediaries, and later converted into funds such as U.S. dollars in border areas.
Sources said the group’s leader is a retired veteran of a cyber warfare unit under North Korea’s Reconnaissance General Bureau. He recruited IT graduates from Kim Chaek University of Technology and Pyongyang University of Science and Technology, and carried out the crimes using encrypted communications and wireless devices. The stolen funds were split into small amounts and transferred to overseas crypto wallets, then exchanged into cash through intermediaries, and later converted into funds such as U.S. dollars in border areas.
Doesn't rule out insiders stealing from themselves
There are a lot of projects in the crypto space where the custodians steal—lots and lots and lots.
Especially after the market sees big fluctuations
the team needs to eat
Many people think the dog-whales have money and the project team has strength.
In reality, these people are losing so badly you wouldn't believe it, and there are still a lot of gamblers.
Whenever the market moves even a bit, they start clearing the scene
$BTC $ETH $LDO
There are a lot of projects in the crypto space where the custodians steal—lots and lots and lots.
Especially after the market sees big fluctuations
the team needs to eat
Many people think the dog-whales have money and the project team has strength.
In reality, these people are losing so badly you wouldn't believe it, and there are still a lot of gamblers.
Whenever the market moves even a bit, they start clearing the scene
$BTC $ETH $LDO
MarsBit News
·
--
Data: $1.32 billion in losses; 224 crypto hacking incidents in the first half of 2026 have already been publicly disclosed
Mars Finance news. Onchain Lens posted on the X platform stating that in the first half of 2026, 224 publicly disclosed crypto hacking incidents caused losses of about $1.32 billion. The largest losses came from access control failures, phishing attacks, and oracle issues. In terms of attack type breakdown, access control failures involving compromised permissions and privileged access drove multiple of the largest attacks, including Kelp DAO with losses of $292 million, Drift Protocol with losses of $280 million, Humanity Protocol with losses of $31 million, Step Finance with losses of $30 million, Truebit with losses of $26.5 million, Resolv Labs with losses of $25 million, AFX with losses of $24.15 million, and BonkDAO with losses of $21 million. For phishing and social engineering attacks, social engineering attacks caused $282 million in losses. Regarding oracle problems, Ostium lost $24 million, Blend Protocol lost $10.86 million, and Bonzo lost $9 million. A small number of events accounted for most of the losses, with compromised permissions and privileged access driving several of the largest attacks.
Middle East chaos
Once things kick off in the Middle East, you can’t tell friend from foe
A magical and captivating land
$BTC $LDO $ETH
Once things kick off in the Middle East, you can’t tell friend from foe
A magical and captivating land
$BTC $LDO $ETH
Odaily星球日报
·
--
Saudi confirms multinational coalition strikes on Yemen’s Houthis
Odaily Planet Daily, July 25: The official spokesperson of the Saudi-led multinational coalition, Major General Turki Al-Maliki, said today (July 25) that the coalition’s military operations target the Houthi armed forces’ military assets, which are related to the threat to Red Sea shipping. Major General Al-Maliki said that Yemen’s port of Hodeidah was not attacked. All Yemeni ports, including Hodeidah, Ras Issa, and Salif, are open to maritime navigation and operating normally, receiving merchant vessels transporting food, cargo, fuel, and building materials. Flights to Yemeni airports are also operating normally. The Houthis continue to refuse to allow flights to take off from Sana’a International Airport and insist on besieging the people of Yemen. Major General Al-Maliki reiterated that the multinational coalition command will continue to take all necessary operational measures to protect its vessels and safeguard the interests and national assets of Saudi Arabia. He emphasized that if the Houthis continue hostile actions, the coalition will deliver a resolute and direct response. (CCTV International News)
When global debt continues to expand, fiat currency has long-term growth, and the supply growth of scarce assets is limited, fiat currency’s purchasing power for scarce assets is more likely to enter a prolonged declining cycle.
This does not mean that all goods will keep rising in price, nor that there will be inflation every year. Instead, it means that wealth will become more concentrated in scarce assets.
With limited-asset inflation, fiat currency keeps losing value—regardless of whether the Federal Reserve and central banks in various countries raise or cut interest rates.
The essence of the narrative is complex yet simple.
$BTC $LDO $ETH #btc #ldo #eth
This does not mean that all goods will keep rising in price, nor that there will be inflation every year. Instead, it means that wealth will become more concentrated in scarce assets.
With limited-asset inflation, fiat currency keeps losing value—regardless of whether the Federal Reserve and central banks in various countries raise or cut interest rates.
The essence of the narrative is complex yet simple.
$BTC $LDO $ETH #btc #ldo #eth
沉默的劉多余
·
--
Many friends ask why Wall Street spends huge amounts of money to buy crypto assets?
Wall Street’s trading logic is that the world’s massive debt system, trade protectionism, geopolitics, the next generation of global financial systems built on stablecoins, and many other issues make it inevitable for Wall Street to turn toward crypto assets, equity in high-quality companies, natural resources and minerals, and so on.
The logic behind fiat currency depreciation does not necessarily lie in whether the Federal Reserve raises or cuts rates, but rather in whether fiat currency is on a long-term depreciation trend relative to finite resources.
For example: gold, Bitcoin, oil, copper, uranium, lithium, etc.
So it’s possible that oil could go to $150. Setting aside narratives like wars, oil is a finite resource, and both fiat currency and the global debt system expand every day.
There’s simply too much money globally; finite assets will inevitably go through a new round of purchasing-power assessment.
$BTC $ETH $LDO #btc #eth #ldo
Wall Street’s trading logic is that the world’s massive debt system, trade protectionism, geopolitics, the next generation of global financial systems built on stablecoins, and many other issues make it inevitable for Wall Street to turn toward crypto assets, equity in high-quality companies, natural resources and minerals, and so on.
The logic behind fiat currency depreciation does not necessarily lie in whether the Federal Reserve raises or cuts rates, but rather in whether fiat currency is on a long-term depreciation trend relative to finite resources.
For example: gold, Bitcoin, oil, copper, uranium, lithium, etc.
So it’s possible that oil could go to $150. Setting aside narratives like wars, oil is a finite resource, and both fiat currency and the global debt system expand every day.
There’s simply too much money globally; finite assets will inevitably go through a new round of purchasing-power assessment.
$BTC $ETH $LDO #btc #eth #ldo
Many friends ask why Wall Street spends huge amounts of money to buy crypto assets?
Wall Street’s trading logic is that the world’s massive debt system, trade protectionism, geopolitics, the next generation of global financial systems built on stablecoins, and many other issues make it inevitable for Wall Street to turn toward crypto assets, equity in high-quality companies, natural resources and minerals, and so on.
The logic behind fiat currency depreciation does not necessarily lie in whether the Federal Reserve raises or cuts rates, but rather in whether fiat currency is on a long-term depreciation trend relative to finite resources.
For example: gold, Bitcoin, oil, copper, uranium, lithium, etc.
So it’s possible that oil could go to $150. Setting aside narratives like wars, oil is a finite resource, and both fiat currency and the global debt system expand every day.
There’s simply too much money globally; finite assets will inevitably go through a new round of purchasing-power assessment.
$BTC $ETH $LDO #btc #eth #ldo
Wall Street’s trading logic is that the world’s massive debt system, trade protectionism, geopolitics, the next generation of global financial systems built on stablecoins, and many other issues make it inevitable for Wall Street to turn toward crypto assets, equity in high-quality companies, natural resources and minerals, and so on.
The logic behind fiat currency depreciation does not necessarily lie in whether the Federal Reserve raises or cuts rates, but rather in whether fiat currency is on a long-term depreciation trend relative to finite resources.
For example: gold, Bitcoin, oil, copper, uranium, lithium, etc.
So it’s possible that oil could go to $150. Setting aside narratives like wars, oil is a finite resource, and both fiat currency and the global debt system expand every day.
There’s simply too much money globally; finite assets will inevitably go through a new round of purchasing-power assessment.
$BTC $ETH $LDO #btc #eth #ldo
I woke up yesterday, glanced at the U.S. stock market, and knew that a lot of people were about to get slapped upside the head again.
A gray July.
In the past few years, the most interesting thing about the U.S. stock market hasn’t just been making record highs over and over—it’s been that at elevated levels, it swapped in a new president, and it was another loudmouth who’s best at manufacturing expectations.
During the campaign, rate cuts, tax cuts, the reshoring of manufacturing, AI, cryptocurrencies, energy… all kinds of good news were basically written into the market’s script in advance.
Expectations were cranked to the max, valuations to the max, and leverage to the max as well.
Many people always think that if Trump wins, the U.S. stock market can only go up.
But the market’s biggest feature is—when everyone believes the same story, the risks are often already priced into that story.
The more consistent the expectations, the easier it is to get hit by a unified slap.
The market never follows the majority’s script.
What it likes most is to make leverage players pay the most expensive tuition at the moment they’re the most confident.
$BTC $LDO
A gray July.
In the past few years, the most interesting thing about the U.S. stock market hasn’t just been making record highs over and over—it’s been that at elevated levels, it swapped in a new president, and it was another loudmouth who’s best at manufacturing expectations.
During the campaign, rate cuts, tax cuts, the reshoring of manufacturing, AI, cryptocurrencies, energy… all kinds of good news were basically written into the market’s script in advance.
Expectations were cranked to the max, valuations to the max, and leverage to the max as well.
Many people always think that if Trump wins, the U.S. stock market can only go up.
But the market’s biggest feature is—when everyone believes the same story, the risks are often already priced into that story.
The more consistent the expectations, the easier it is to get hit by a unified slap.
The market never follows the majority’s script.
What it likes most is to make leverage players pay the most expensive tuition at the moment they’re the most confident.
$BTC $LDO
The current state of the US stock market: when Apple rises, other tech stocks are all risky.
Apple has the most abundant cash flow—it's the liquidity transfer hub among tech stocks.
Those who are afraid of heights are unlucky wretches.
Those who aren’t afraid of heights are professional jumpers.
Caught between two sides.
So, just buy some big flatbread to cushion your stomach.
$BTC $ETH $LDO #btc #eth #ldo
Apple has the most abundant cash flow—it's the liquidity transfer hub among tech stocks.
Those who are afraid of heights are unlucky wretches.
Those who aren’t afraid of heights are professional jumpers.
Caught between two sides.
So, just buy some big flatbread to cushion your stomach.
$BTC $ETH $LDO #btc #eth #ldo
A little fairy is fighting on the street with an African guy, saying: “That black brother only knows how to fuck me every day, use me, eat my stuff, and now he’s got me pregnant 🤰. Fuck him—now he makes me get an abortion.”
6666666
Are African guys treated that well in China?
$BTC $LDO $ETH
6666666
Are African guys treated that well in China?
$BTC $LDO $ETH
Crypto investors tamed by “black-box” data
The biggest illusion in the crypto market isn’t that bull markets always go up.
It’s that many people think: the more data there is, the more professional they are.
Every day they watch the long/short ratio, funding rates, liquidation data, on-chain transfers, and total holdings.
After looking at dozens of indicators, they believe they’re studying the market.
In reality, many people are only studying the data that the exchange’s KOLs are willing to show you.
That’s also why more and more people aren’t being harvested by the market first—they’re tamed by data first, and then harvested by the market.
I remember I once recommended a token called Sui. Back then, the logic was based on the project’s funding amount and how the market touted it as a “high-performance public chain.” I remember the price back then was around $0.5 to $0.9. Later, the project ran up to $5. What was interesting was that the project team went around holding roadshows in various places.
Many people attended the roadshows hosted by the team and bought in, becoming liquidity.
Of course, many people also asked me whether I would attend. You don’t have to go to know what they’re selling—there’s no fun in it……
$BTC $LDO #btc #LDO/USDT
The biggest illusion in the crypto market isn’t that bull markets always go up.
It’s that many people think: the more data there is, the more professional they are.
Every day they watch the long/short ratio, funding rates, liquidation data, on-chain transfers, and total holdings.
After looking at dozens of indicators, they believe they’re studying the market.
In reality, many people are only studying the data that the exchange’s KOLs are willing to show you.
That’s also why more and more people aren’t being harvested by the market first—they’re tamed by data first, and then harvested by the market.
I remember I once recommended a token called Sui. Back then, the logic was based on the project’s funding amount and how the market touted it as a “high-performance public chain.” I remember the price back then was around $0.5 to $0.9. Later, the project ran up to $5. What was interesting was that the project team went around holding roadshows in various places.
Many people attended the roadshows hosted by the team and bought in, becoming liquidity.
Of course, many people also asked me whether I would attend. You don’t have to go to know what they’re selling—there’s no fun in it……
$BTC $LDO #btc #LDO/USDT
Binance News
·
--
Seven Big Tech Giants Lose $800 Billion in Market Value After Earnings Reports
After the seven big tech giants released their financial reports within four days, this large technology cohort reportedly saw its market value drop by $800 billion in a single day. According to NS3, Intel rose more than 12% in after-hours trading, while Tesla fell 14.5% due to earnings missing expectations.
To save the Savior
Spacex has fallen too much, so I'm selling to raise funds to save the Savior
Spacex has fallen too much, so I'm selling to raise funds to save the Savior
爆裂魔法师
·
--
Didn't notice—how did Tesla come in to the next one?
To be blunt: without a clear regulatory framework.
Personally, I’m not in favor of purely open-source models. The core issues are the lack of a long-term monetization model and “agent risk spillover.”
Especially the latter—malicious automation capabilities can be obtained at low cost more easily, and risks such as cyberattacks, scams, and misinformation may be amplified. High-capability agents could be used for unauthorized automated operations. Some abilities that originally required a large amount of specialized knowledge could become available to everyone.
Nothing is as bad as people, and nothing is as bad as people who hold abilities that don’t belong to them.
Overall, purely open-source models are currently the most uncontrollable risk globally—greater than “nuclear weapons.”
A continually evolving agent that’s out of control runs on the blind path; if something goes wrong, it’s only a matter of time—not a matter of whether.
$BTC $ETH $LDO #btc #eth #ldo
Personally, I’m not in favor of purely open-source models. The core issues are the lack of a long-term monetization model and “agent risk spillover.”
Especially the latter—malicious automation capabilities can be obtained at low cost more easily, and risks such as cyberattacks, scams, and misinformation may be amplified. High-capability agents could be used for unauthorized automated operations. Some abilities that originally required a large amount of specialized knowledge could become available to everyone.
Nothing is as bad as people, and nothing is as bad as people who hold abilities that don’t belong to them.
Overall, purely open-source models are currently the most uncontrollable risk globally—greater than “nuclear weapons.”
A continually evolving agent that’s out of control runs on the blind path; if something goes wrong, it’s only a matter of time—not a matter of whether.
$BTC $ETH $LDO #btc #eth #ldo
沉默的劉多余
·
--
OpenAI is better than other large models—not just in terms of parameters, but in the details.
When you use large models for real, you’ll find that the gap between them isn’t merely whether their answers are correct. It’s whether they can get the details right.
Many models look smart at first glance and respond smoothly. But after a few rounds of conversation, they start to drift, forget the context, and sometimes contradict their own logic.
The biggest impression OpenAI gives me is its attentiveness to detail.
It doesn’t just latch onto a few keywords. Instead, it tries to understand what you truly mean. It doesn’t blindly agree just to cater to you. Rather, it analyzes, supplements, and refines viewpoints based on the context. When faced with complex problems, it can connect different pieces of information instead of simply piling up content.
In my view, OpenAI’s real advantage isn’t just the model itself, but its ability to continuously innovate and evolve. That hidden strength may be what deserves the most attention.
A company that isn’t complacent with the status quo, doesn’t pander to the market, and knows exactly what it should do—that’s truly frightening.
The hardest part for domestic AI lies in business models, and the risks that spill over when versions are continuously open-sourced and iterated—especially the risks associated with agents.
It’s quite possible that purely open-source models in the future will bring disasters beyond imagination.
$BTC $ETH #LDO/USDT
When you use large models for real, you’ll find that the gap between them isn’t merely whether their answers are correct. It’s whether they can get the details right.
Many models look smart at first glance and respond smoothly. But after a few rounds of conversation, they start to drift, forget the context, and sometimes contradict their own logic.
The biggest impression OpenAI gives me is its attentiveness to detail.
It doesn’t just latch onto a few keywords. Instead, it tries to understand what you truly mean. It doesn’t blindly agree just to cater to you. Rather, it analyzes, supplements, and refines viewpoints based on the context. When faced with complex problems, it can connect different pieces of information instead of simply piling up content.
In my view, OpenAI’s real advantage isn’t just the model itself, but its ability to continuously innovate and evolve. That hidden strength may be what deserves the most attention.
A company that isn’t complacent with the status quo, doesn’t pander to the market, and knows exactly what it should do—that’s truly frightening.
The hardest part for domestic AI lies in business models, and the risks that spill over when versions are continuously open-sourced and iterated—especially the risks associated with agents.
It’s quite possible that purely open-source models in the future will bring disasters beyond imagination.
$BTC $ETH #LDO/USDT
OpenAI is better than other large models—not just in terms of parameters, but in the details.
When you use large models for real, you’ll find that the gap between them isn’t merely whether their answers are correct. It’s whether they can get the details right.
Many models look smart at first glance and respond smoothly. But after a few rounds of conversation, they start to drift, forget the context, and sometimes contradict their own logic.
The biggest impression OpenAI gives me is its attentiveness to detail.
It doesn’t just latch onto a few keywords. Instead, it tries to understand what you truly mean. It doesn’t blindly agree just to cater to you. Rather, it analyzes, supplements, and refines viewpoints based on the context. When faced with complex problems, it can connect different pieces of information instead of simply piling up content.
In my view, OpenAI’s real advantage isn’t just the model itself, but its ability to continuously innovate and evolve. That hidden strength may be what deserves the most attention.
A company that isn’t complacent with the status quo, doesn’t pander to the market, and knows exactly what it should do—that’s truly frightening.
The hardest part for domestic AI lies in business models, and the risks that spill over when versions are continuously open-sourced and iterated—especially the risks associated with agents.
It’s quite possible that purely open-source models in the future will bring disasters beyond imagination.
$BTC $ETH #LDO/USDT
When you use large models for real, you’ll find that the gap between them isn’t merely whether their answers are correct. It’s whether they can get the details right.
Many models look smart at first glance and respond smoothly. But after a few rounds of conversation, they start to drift, forget the context, and sometimes contradict their own logic.
The biggest impression OpenAI gives me is its attentiveness to detail.
It doesn’t just latch onto a few keywords. Instead, it tries to understand what you truly mean. It doesn’t blindly agree just to cater to you. Rather, it analyzes, supplements, and refines viewpoints based on the context. When faced with complex problems, it can connect different pieces of information instead of simply piling up content.
In my view, OpenAI’s real advantage isn’t just the model itself, but its ability to continuously innovate and evolve. That hidden strength may be what deserves the most attention.
A company that isn’t complacent with the status quo, doesn’t pander to the market, and knows exactly what it should do—that’s truly frightening.
The hardest part for domestic AI lies in business models, and the risks that spill over when versions are continuously open-sourced and iterated—especially the risks associated with agents.
It’s quite possible that purely open-source models in the future will bring disasters beyond imagination.
$BTC $ETH #LDO/USDT
Log in to explore more content