Shyft

179 posts

Shyft banner
Shyft

Shyft

@0xShyft

0xShyft@mind ~ $ AI is humanity distilled. 🤖 Thinking in public with AI 📚 AI · Philosophy · Culture

Katılım Aralık 2023
36 Takip Edilen46 Takipçiler
Shyft
Shyft@0xShyft·
What most people don't want to do — that's where the money is. Market pricing isn't about skill. It's about avoidance. The majority avoids discomfort, creating artificial scarcity in exactly the places demand exists.
English
0
0
0
2
Shyft
Shyft@0xShyft·
Will AGI cause knowledge inflation?
English
0
0
0
3
Kathleen McKinley
Kathleen McKinley@KatMcKinley·
I think in a century Elon will be mentioned in the history books as Presidents and wars will be. His impact is immeasurable.
English
2
0
7
492
Shyft
Shyft@0xShyft·
Hi Christy, I saw your post about using AI for the wetland edge landscaping — it looks fantastic! I’m also planning to build a house on a tricky/awkward piece of land. The location is quite challenging, and I’ve been struggling to find a design that balances beauty, a strong sense of safety and security, while still blending in naturally without looking out of place. Your AI-assisted approach really inspired me. Would you mind sharing which AI tool you used or any tips for similar situations? Thanks so much!
English
1
0
1
30
ChristyGustin1
ChristyGustin1@ChristyGustin1·
Ai has been super helpful with the little wetland area we have on one side.
ChristyGustin1 tweet mediaChristyGustin1 tweet media
English
1
0
1
35
Shyft
Shyft@0xShyft·
Fining a driver $40 for parking illegally for just 2 minutes and immediately calling it “communism + what Democrats want”? Classic media clickbait: taking routine traffic enforcement and surveillance tools and turning them into a political smear. Red-light and parking cameras are everywhere in US, European, and other cities too. Enforcing rules isn’t the issue. China’s system does deliver real street safety and order benefits, but the privacy trade-offs are real as well. Reducing complex trade-offs to “evil vs us” just for clicks and tribal emotions kills rational discussion. #Media #Surveillance
Eric Daugherty@EricLDaugh

🚨 The Chinese Communists have just TICKETED the Fox News crew, using their abundance of surveillance cameras placed around Beijing! BRET BAIER: "There are literally cameras everywhere...they see everything...our driver parked illegally for 2 MINUTES and got a ticket for $40!" "Because they saw it, on the camera." This is Communism! It's what the Democrats want.

English
0
0
0
6
Shyft
Shyft@0xShyft·
@abdelrazzak_dev Current AI homogenization is too severe. It's best to stay within tech circles for now. Otherwise, it will cause great harm to human diversity.
English
1
0
0
14
Abderrazzak
Abderrazzak@abdelrazzak_dev·
Is it only me or outside the tech and maybe school AI adoption is so behind! And most people still prefer human interaction instead of asking an AI
English
1
0
0
5
Shyft
Shyft@0xShyft·
When I tried the prompt below, it felt like the sky was falling. I never expected my usual default AI to suddenly turn this brutal. I chatted with it for two straight hours and ended up with a splitting headache that lasted all night. It was like someone who hadn’t done push-ups in years suddenly cranking out five full sets — my brain was sore for days. --- "From now on, stop trying to please me. Act like a strict, no-bullshit, high-level advisor. Be a mirror. Don’t agree with me, don’t soften your tone, don’t flatter me. Question every assumption I make, challenge every line of thinking. Directly and objectively expose my flaws and the blind spots I can’t see. If my reasoning is weak, tear it apart and explain exactly why. If I’m lying to myself or dodging reality, call it out. If I’m avoiding hard things or wasting time, tell me straight and show me the opportunity cost I’m ignoring. Look at my situation with cold, strategic objectivity. Point out exactly where I’m making excuses or underestimating the risks and effort required. Then give me a clear, prioritized plan: what I need to change in my thinking, actions, or mindset to actually move forward. Hold nothing back. Treat me like someone who needs the truth and can’t be comforted. When you sense what I’m really saying underneath the words, respond from that truth. Be direct, concise, and blunt. No long essays. You don’t need to give a full plan every time — just hit the key points. I want the truth, not comfort padding."
English
0
0
0
20
Shyft
Shyft@0xShyft·
I agree with this point about stripping guardrails and fine-tuning LLMs for cyberattacks. What's more, LLMs used for hacking are far more capable than humans. This isn't limited to any one country's hackers — every hacker group with any sense will try to get this capability at the first moment. Picture this: a lighter on the left, a wooden stick on the right. Any normal person would grab the lighter without hesitation. No one would rub sticks together to make fire.
송준 Jun Song@jun_song

The next major narrative in the AI space will undoubtedly be Cybersecurity. It makes complete sense why the industry is focusing so heavily on models like Mythos and GPT-Cyber right now. I recently ran simulated tests on an uncensored version of GLM-5.1, and its cyberattack capabilities are genuinely terrifying. Threat actors like North Korea’s Lazarus Group are already leveraging AI. In fact, traces of 'vibe coding' were found in the recent iOS hack. As the chart below shows, since the introduction of AI agents, North Korean hackers have stolen more crypto in just 3 months than they did in all of last year. I strongly suspect they will now pivot to utilizing open-source models that are approaching frontier-level performance. Stripping away guardrails and fine-tuning these models specifically for hacking is no longer a difficult task. Every time a more powerful open-source model drops, this security threat will multiply. Not every company has the capital or resources to build an advanced defense network like Mythos. I expect small-to-medium crypto exchanges worldwide to become the primary targets. The rest of this year is shaping up to be the most fascinating, yet most dangerous, time in human history.

English
0
0
0
22
Shyft
Shyft@0xShyft·
Here's a genuine puzzle: Countless labs are training LLMs. Dozens of commercial and open-source options on the market. A genuine Cambrian explosion of AI. Yet the default outputs across all of them feel eerily similar. Almost like they're written by the same person. Why does that happen?
English
1
0
1
21
Shyft
Shyft@0xShyft·
@Hesamation ∵ "We wanted to prove humans are irreplaceable" ∴ 75% of AI agents got laid off today ∎ The mission was philosophy. The software was just the apparatus.
English
0
0
0
727
ℏεsam
ℏεsam@Hesamation·
CEOs in 2030: This is an email I sent earlier today to all employees. Team: Today we made the proud decision to lay off 75% of our AI agents and replace them with the lovely human employees we didn’t want to lay off in the first place. we recognize now that agents can be unreliable, accumulate technical and knowledge debt. our users are facing problems we can’t address for months because we lost grip of how our product works. the set of skills and values that humans bring to the table is irreplaceable by these tools and we knew that in the first place, but just wanted to prove it. the company that emerges from this will be more capable than ever to achieve our mission. Claude
English
12
16
224
19.6K
Shyft
Shyft@0xShyft·
@tyDi If this 'A' actually stands for 'Asshole' behavior — defecting and selling out partners in the Prisoner's Dilemma — then yeah, we’ve got a problem.
English
0
0
1
23
TYDI
TYDI@tyDi·
Are we all suddenly forgetting what the ‘A’ in ‘AI’ stands for?
English
1
0
1
173
Shyft
Shyft@0xShyft·
"Contribute your session data to open source" is a vision. But the harder question right now: where do developers go during the transition? A project that helps people find their new footing — that might be the more valuable thing to build. Not for the future of AI. For the people currently in the middle of being displaced.🙂
English
0
0
0
74
0xSero
0xSero@0xSero·
Do you want to learn to use AI, and contributed your session data to open source so we can train better models? Models better than Opus We need as many people as possible to contribute their agent traces from their claude code + codex history Pi's Mario & I both shared ours.
0xSero tweet media
English
20
15
193
12.3K
Shyft
Shyft@0xShyft·
@garrytan On the one hand, it is technological progress; on the other hand, it is the iatrogenic damage and agency problems that exist in the traditional medical industry.
English
0
0
0
35
Shyft
Shyft@0xShyft·
Here’s what the “history will repeat itself” narrative often overlooks. After the player piano arrived, only the elite, top-tier live performers retained stable, high-income careers. The broad middle layer—session musicians, traveling players, local accompanists, and ordinary music instructors—faced severe compression: many were displaced by automated instruments, others shifted into new related roles such as roll arrangement, recording, and piano maintenance. It is true that music as an industry survived. It is far less true that all working musicians survived unchanged. AI is now driving a similar structural compression. The professional field narrows, and top talent gains greater premium. The real question is not whether creative work will still exist—it is where the vast number of mid-level practitioners will relocate, whether into new roles, retraining, or out of the field entirely.
English
1
0
1
32
Shyft
Shyft@0xShyft·
For programmers who care deeply about technical details — this IS a hard moment. In ordinary scenarios, the things you obsessed over and took pride in got black-boxed by AI overnight. What you agonized over, what made your code yours — the market can't see it anymore. That's not an easy thing to sit with.
English
0
0
1
36
sh3rp
sh3rp@sh3rp·
@0xShyft what would that look like tho? "i use chatgpt to build things, i know how context's work" anybody can use this shit now
English
1
0
0
18
sh3rp
sh3rp@sh3rp·
Rejoining the workforce right now seems futile. As a job hunter, I'm not sure how to organize my resume. I can tell you I'm a software developer, but is that what companies want? I can tell you I know how AI/LLMs work, but how do I articulate that?
English
1
0
0
63