en
Feedback
Continuous Learning_Startup & Investment

Continuous Learning_Startup & Investment

Open in Telegram

We journey together through the captivating realms of entrepreneurship, investment, life, and technology. This is my chronicle of exploration, where I capture and share the lessons that shape our world. Join us and let's never stop learning!

Show more
2 287
Subscribers
-324 hours
-167 days
-4530 days
Posts Archive
next up is @DedyKredo LIVE CODING a full test suite, making code changes, and automating commit and PR review, all assisted by @CodiumAI . audible β€œwhat the fuck” from @eugeneyan. youtube.com/live/qw4PrtyvJ… ends with a powerful message for Israel. we stand with you @itamar_mar.

it’s official - I think GitHub Copilot is the first* generative AI product to publicly claim they’ve passed $100m ARR β€” enough to stand alone as a publicly listed company Whenever people ask me β€œis AI a fad” the biggest thing I point to is β€œfollow the money”: - revenue, not just funding - RECURRING, not tcosts on hype - people publicly saying they’d pay 5x the cost (*there’s likely a few others but none confirmed officially - see Anatomy of Autonomy post on @latentspacepod)

β€œThe question we always focus on is β€˜can this company become a monopoly?’” He then lists several things that can make a company a monopoly: Super fast distribution on a very thin product (e.g. Twitter) A technological advantage that is continually built upon: you come up with something new and steadily improve (e.g. enterprise SaaS software) A truly brilliant breakthrough (e.g. Bitcoin) However he argues that complex coordinationβ€”where you take a lot of little pieces and coordinate them into something newβ€”is continually overlooked as a way to create a monopoly: β€œThis is the thing that’s maybe 180 degrees antithetical to the Lean Startup ethos. It’s complicated. You have to put all the pieces together in just the right way. I think this is on some level what really drove Apple as an innovative company in the last decade… What was new about the iPhone? There was no single component that was new. It was just that you put all of these things together in just the right way… and once you built it, it was actually super hard for people to replicate. You had an advantage for many years. You could get network lock inβ€”in terms of the app community or the brand.” He also points to Tesla and SpaceX as examples: β€œThere’s no component to the Tesla that’s actually that new. It’s just that you put all of the pieces together. You re-engineered the whole distributor network. It was this complex coordination that made it work. There’s like this lost art of accounting where you figure out how much things cost and add them all together. And Elon has discovered this lost art of accounting which no other people practice.” https://x.com/mikemcg0/status/1711727266537812429?s=46&t=h5Byg6Wosg8MJb4pbPSDow

πŸ€”πŸ˜‚
πŸ€”πŸ˜‚

Punch Cards Era: The early days of computing required users to interact with machines using punch cards. These rectangular pieces of stiff paper had holes punched into them, representing data and commands. It was a labor-intensive process and the room for error was vast. If one card was misplaced, the whole sequence would be thrown off. 2. Command-Line Interfaces (CLI): The 1980s saw a shift from punch cards to command-line interfaces. Computers like the IBM PC and Apple Macintosh popularized the CLI. While it was more efficient than punch cards, it still required users to memorize commands and their syntax to communicate with the computer. 3. Graphical User Interfaces (GUI): As technology progressed, GUIs began to emerge in the late 1980s and early 1990s. The Apple Macintosh and Microsoft Windows operating systems popularized this interface. Icons, windows, and point-and-click mechanisms made computing more accessible and intuitive for the masses. 4. Touch Interfaces: The 2000s heralded the age of touchscreens. Devices like smartphones and tablets brought a more intimate and direct way of interacting with computers. Pinching, zooming, and swiping became the new language of interaction. 5. Voice Recognition: With the rise of digital assistants like Siri, Alexa, and Google Assistant, voice became a primary mode of interaction. This allowed for hands-free computing and made technology even more embedded in our daily lives. 6. Generative AI and Conversational Interfaces: Today, we're in the age of conversational AI, epitomized by platforms like ChatGPT. These systems not only understand human language but can also generate human-like responses. It feels less like communicating with a machine and more like having a conversation with another human. https://www.linkedin.com/pulse/from-punch-cards-conversational-ai-evolution-computer-adriana-rocha-1f

20 years in building companies taught me: speed is king. Moving fast is a miracle drug. Here's why: β†’ You learn more about the end-state product per unit time. There is no team able to accurately predict every future product need. Having an iterative product schedule will solve for this. β†’ Achieves a more robust product. If I had to summarize technology development: it's how many iterations you have done and then how much progress you've made b/t those iterations. β†’ Helps prioritize what's important. Speeds means you only have time for the priority matters. There is no time for things that don't matter. β†’ Time is what will kill your company. P.S. Moving fast is so important to me that it's a corporate value at Figure.

λ©”νƒ€λ²„μŠ€κ°€ μŠ€λ§ˆνŠΈν°μ„ λŒ€μ²΄ν•  νΌνŒ©ν„°λ‘œμ„œ μ—¬μ „νžˆ λ¬΄ν•œν•œ 잠재λ ₯을 κ°€μ§€κ³  μžˆμ§€λ§Œ κ°€μž₯ 큰 이슈인 νœ΄λŒ€μ„±μ„ ν•΄κ²°ν•˜μ§€ λͺ»ν•  경우 μ—¬μ „νžˆ λŒ€μ€‘ν™”λŠ” μš”μ›ν•˜λ‹€λŠ” ν‰κ°€μž…λ‹ˆλ‹€.Β  μ•„μ΄ν°μ˜ λ“±μž₯κ³Ό ν•¨κ»˜ μ‹œμž‘λœ 슀마트폰의 λŒ€μ€‘ν™” μ΄μ „μ—λŠ” 무렀 10년에 걸쳐 λΈ”λž™λ² λ¦¬, 팜 파일럿, μœˆλ„μš°ν°κ³Ό 같은 λ‹€μ–‘ν•œ μ‹œλ„κ°€ 이어지며 μ‹œν–‰μ°©μ˜€λ₯Ό κ²ͺ은 λ°” μžˆμŠ΅λ‹ˆλ‹€. 슀마트폰의 λ°œμ „ 과정에 λΉ„μΆ°λ³Ό λ•Œ λ©”νƒ€λ²„μŠ€λŠ” 아직 '아이폰 λͺ¨λ¨ΌνŠΈ'λŠ” 컀녕 'λΈ”λž™λ² λ¦¬ λͺ¨λ¨ΌνŠΈ'에도 λ„λ‹¬ν•˜μ§€ λͺ»ν–ˆλ‹€κ³  것이 λƒ‰μ •ν•œ ν‰κ°€μž…λ‹ˆλ‹€. λ©”νƒ€ν€˜μŠ€νŠΈ 3와 μ• ν”Œ λΉ„μ „ ν”„λ‘œκ°€ μ΄λŸ¬ν•œ 뢀정적인 여둠을 잠재우고 λ°˜μ „μ„ 이뀄낼 수 μžˆμ„μ§€ κ·€μΆ”κ°€ μ£Όλͺ©λ©λ‹ˆλ‹€.Β 

<머슀크의 미래λ₯Ό μ˜ˆμƒν•΄λ³΄μž> 책이 νŠΈμœ„ν„° 인수λ₯Ό 포함해 μ΅œμ‹  μ΄μŠˆλ“€κΉŒμ§€ ν¬ν•¨ν•˜κ³  μžˆμ–΄μ„œ 머슀크의 μ—¬λŸ¬ μ‚¬μ—…μ²΄μ˜ λ―Έλž˜μ— λŒ€ν•œ 힌트λ₯Ό μ–»λŠ”λ°λ„ μœ μš©ν–ˆλ‹€. μš°μ„  μ§€κΈˆμ€ μ—‘μŠ€κ°€ λ˜μ–΄λ²„λ¦° νŠΈμœ„ν„°. 책을 μ½μ–΄λ³΄λ‹ˆ νŠΈμœ„ν„°λŠ” 머슀크 특유의 Surgeκ°€ λ°œλ™ν•΄μ„œ μΆ©λ™μ μœΌλ‘œ μΈμˆ˜ν•œ κ²ƒμœΌλ‘œ λ³΄μ΄λŠ”λ°, λ¨ΈμŠ€ν¬λŠ” ν•œ λ•Œ μ—‘μŠ€λ‹·μ»΄(νŽ˜μ΄νŒ”)의 κ²½μ˜μžμ˜€λ‹€λŠ”κ±Έ μžŠμ–΄μ„œλŠ” μ•ˆλœλ‹€. κ·ΈλŠ” νŠΈμœ„ν„°λ₯Ό μ†Œμ…œ λ„€νŠΈμ›Œν¬μ™€ 결제 ν”Œλž«νΌμ΄ κ²°ν•©λœ μ„œλΉ„μŠ€λ‘œ λ§Œλ“€ κ³„νšμ΄λ‹€. "λ¨ΈμŠ€ν¬κ°€ κ΅¬μƒν•œ μ—‘μŠ€λ‹·μ»΄μ˜ μ½˜μ…‰νŠΈλŠ” μ›λŒ€ν–ˆλ‹€. λ±…ν‚Ήκ³Ό λ””μ§€ν„Έ ꡬ맀, λ‹Ήμ’Œμ˜ˆκΈˆ, μ‹ μš©μΉ΄λ“œ, 투자, λŒ€μΆœ λ“± λͺ¨λ“  금육 μ„œλΉ„μŠ€λ₯Ό μ œκ³΅ν•˜λŠ” μ›μŠ€ν†± 온라인 은행을 λ§Œλ“œλŠ” κ²ƒμ΄μ—ˆλ‹€. κ±°λž˜λŠ” κ²°μ œκ°€ μ™„λ£Œλ  λ•ŒκΉŒμ§€ 기닀릴 ν•„μš” 없이 μ¦‰μ‹œ μ²˜λ¦¬λ˜λŠ” λ°©μ‹μ΄μ—ˆλ‹€. λ¨ΈμŠ€ν¬λŠ” 돈이 λ°μ΄ν„°λ² μ΄μŠ€μ— μž…λ ₯λ˜λŠ” ν•­λͺ©μ— λΆˆκ³Όν•˜λ‹€λŠ” 톡찰을 λ°”νƒ•μœΌλ‘œ, λͺ¨λ“  거래λ₯Ό μ‹€μ‹œκ°„μœΌλ‘œ μ•ˆμ „ν•˜κ²Œ κΈ°λ‘ν•˜λŠ” 방법을 κ³ μ•ˆν•˜κ³  μ‹Άμ—ˆλ‹€. β€œμ†ŒλΉ„μžκ°€ μ‹œμŠ€ν…œμ—μ„œ λˆμ„ μΈμΆœν•˜λŠ” λͺ¨λ“  이유λ₯Ό ν•΄κ²°ν•΄μ€€λ‹€λ©΄, λͺ¨λ“  돈이 λͺ¨μ΄κ²Œ 될 것이고, κ·Έλ ‡κ²Œ 되면 수쑰 λ‹¬λŸ¬ 규λͺ¨μ˜ νšŒμ‚¬κ°€ 될 수 μžˆμ„κ±°λΌκ³  μƒκ°ν–ˆμ–΄μš”.” 머슀크의 μ„€λͺ…이닀." 그런데 졜근 λ“€μ–΄μ„œλŠ” νŠΈμœ„ν„°μ— λŒ€ν•œ ν₯λ―Έκ°€ 전보닀 μ€„μ–΄λ“€μ—ˆκ³ , 인곡지λŠ₯에 λŒ€ν•œ 관심이 더 컀진 μƒνƒœμΈ 것 κ°™λ‹€. β€œμΈκ³΅μ§€λŠ₯κ³Ό κ΄€λ ¨λœ 상황을 κ³ λ €ν•  λ•Œ νŠΈμœ„ν„°μ— λŒ€ν•΄ κ·Έλ ‡κ²Œ λ§Žμ€ μ‹œκ°„μ„ ν• μ• ν•  κ°€μΉ˜κ°€ μžˆλŠ”μ§€ 의문이 λ“­λ‹ˆλ‹€. λ¬Όλ‘  νŠΈμœ„ν„°λ₯Ό μ„Έκ³„μ—μ„œ κ°€μž₯ 큰 κΈˆμœ΅κΈ°κ΄€μœΌλ‘œ λ§Œλ“€ 수 μžˆκ² μ§€μš”. ν•˜μ§€λ§Œ λ‚΄ λ‘λ‡Œ ν™œλ™μ˜ 주기와 ν•˜λ£¨μ˜ μ‹œκ°„μ€ ν•œμ •λ˜μ–΄ μžˆμž–μ•„μš”. 더 λΆ€μžλ‚˜ 뭐 그런 게 λ˜μ–΄μ•Ό ν•˜λŠ” 것도 μ•„λ‹ˆκ³ μš”.” μ—¬λ‹΄μœΌλ‘œ 그의 동생인 ν‚΄λ²Œμ€ μΌλ‘ μ—κ²Œ 블둝체인 기반의 μ†Œμ…œ ν”Œλž«νΌ 아이디어λ₯Ό μ œκ³΅ν–ˆκ³ , λ¨ΈμŠ€ν¬λŠ” 이λ₯Ό ν”ŒλžœB라고 λΆˆλ €λ‹€. λ§Œμ•½ νŠΈμœ„ν„° μΈμˆ˜κ°€ κ²°λ ¬λ˜μ—ˆλ‹€λ©΄ 블둝체인 μ„œλΉ„μŠ€λ₯Ό λ§Œλ“€μ—ˆμ„μ§€λ„? λ¬Όλ‘  λ¨ΈμŠ€ν¬λŠ” νŠΈμœ„ν„° 데이터λ₯Ό μ²˜λ¦¬ν•˜κΈ°μ—” λΈ”λ‘μ²΄μΈμ˜ 속도가 λ„ˆλ¬΄ λŠλ¦¬λ‹€κ³  μƒκ°ν•˜κ³  있긴 ν•˜λ‹€. 그리고 νŠΈμœ„ν„°κ°€ λ³΄μœ ν•œ β€˜λ°μ΄ν„°β€™μ˜ κ°€μΉ˜λŠ” 인수 후에 κΉ¨λ‹¬μ•˜λ‹€κ³  ν•œλ‹€. 즉, 데이터λ₯Ό μœ„ν•΄μ„œ κ·Έ 큰 κΈˆμ•‘μ„ μ§€λ₯Έκ±΄ μ•„λ‹ˆμ—ˆλ‹€λŠ” 의미 (μ§„μ§œλ‘œ κ·Έλƒ₯ 사고 μ‹Άμ–΄μ„œ μ‚° 것에 가깝닀.) μ•„λ¬΄λž˜λ„ 주주인만큼 ν…ŒμŠ¬λΌ 이야기가 κ°€μž₯ ν₯미둭게 μ½νžˆλŠ”λ°, λ¨ΈμŠ€ν¬λŠ” μ²˜μŒμ—λŠ” 2만 5μ²œλ‹¬λŸ¬ 짜리 μžλ™μ°¨ λ§Œλ“œλŠ” 것에 λΆ€μ •μ μ΄μ—ˆλ‹€κ³  ν•œλ‹€. λ‘œλ³΄νƒμ‹œκ°€ κ³§ λŒ€μ€‘ν™”λ˜λ©΄ ν•„μš” 없을 κ²ƒμ΄λΌλŠ” 이유둜. ν•˜μ§€λ§Œ ν…ŒμŠ¬λΌ λ””μžμ΄λ„ˆμΈ ν”„λž€μΈ  폰 ν™€μΈ ν•˜μš°μ  μ΄ 사이버 트럭 λΉ„μŠ·ν•˜κ²Œ 생긴 μ°¨λŸ‰ λͺ¨ν˜•을 λ³΄μ—¬μ£Όλ‹ˆ 생각이 λ°”λ€Œμ—ˆλ‹€κ³  ν•œλ‹€. 이 μ°¨λŸ‰μ—λŠ” μ°¨μ„ΈλŒ€ ν”Œλž«νΌμ΄ 적용될 μ˜ˆμ •μ΄κ³ , μ›λž˜λŠ” μ°¨μ„ΈλŒ€ κΈ°κ°€νŒ©ν† λ¦¬μΈ λ©•μ‹œμ½”μ—μ„œ 생산 μ˜ˆμ •μ΄μ—ˆμ§€λ§Œ μ΅œκ·Όμ— μ˜€μŠ€ν‹΄μœΌλ‘œ λ³€κ²½λ˜μ—ˆλ‹€κ³  ν•œλ‹€. μ΄μœ λŠ” μ—”μ§€λ‹ˆμ–΄λ“€μ„ λ©•μ‹œμ½”λ‘œ μ΄μ£Όμ‹œν‚€λŠ”κ²Œ μ–΄λ ΅κΈ° λ•Œλ¬Έμ—, λΉ λ₯Έ ν”Όλ“œλ°±μ„ μœ„ν•΄μ„œλŠ” 본인 μ§‘κ³Ό κ°€κΉŒμš΄ ν…μ‚¬μŠ€μ—μ„œ μƒμ‚°ν•˜λŠ”κ²Œ λ§žλ‹€κ³  νŒλ‹¨ν–ˆλ‹€κ³ . λ¨ΈμŠ€ν¬λŠ” 이번 여름 λ‚΄λ‚΄ 이λ₯Ό μœ„ν•œ 생산 곡정을 λ°œμ „μ‹œν‚€λŠ”λ° μ‹œκ°„μ„ νˆ¬μžν–ˆλ‹€κ³  ν•œλ‹€. μžμœ¨μ£Όν–‰ 이야기도 μžμ„Έν•˜κ²Œ λ“±μž₯ν•œλ‹€. λ‚˜λŠ” λ¨ΈμŠ€ν¬κ°€ λ ˆμ΄λ”(라이닀)λ₯Ό λ°˜λŒ€ν•˜λŠ”κ²Œ λ‹¨μˆœνžˆ 고집인쀄 μ•Œμ•˜λŠ”λ° 직관적인 μ΄μœ κ°€ μžˆλ”λΌ. λ°”λ‘œ 인간이 μ‹œκ° λ°μ΄ν„°λ§ŒμœΌλ‘œ μš΄μ „μ„ ν•  수 있기 λ•Œλ¬Έμ— 기계도 κ·Έλ ‡κ²Œ ν•  수 μžˆμ–΄μ•Ό ν•œλ‹€λŠ” 이유인데, λ„ˆλ¬΄ λ§žλŠ” 말이닀. 그리고 λ ˆμ΄λ”λ₯Ό μ™„μ „νžˆ λ°˜λŒ€ν•˜λŠ” 것도 μ•„λ‹Œμ˜€λ˜ 것이 슈쑰우의 νŒ€μ΄ λ ˆμ΄λ” μ‹œμŠ€ν…œμ„ λ”°λ‘œ κ°œλ°œν•˜κΈ°λ„ ν–ˆμœΌλ©°, 머슀크 λ˜ν•œ λͺ¨λΈ S와 Y에 λ ˆμ΄λ”λ₯Ό μ‹œν—˜ν•΄λ³΄λŠ”κ±Έ μŠΉμΈν–ˆλ‹€κ³  ν•œλ‹€. β€œμΌλ°˜μ μΈ μžλ™μ°¨ λ ˆμ΄λ”λ³΄λ‹€ 훨씬 더 μ •κ΅ν•œ λ ˆμ΄λ”μ΄μ§€μš”. 무기 μ‹œμŠ€ν…œμ—μ„œ λ³Ό 수 μžˆλŠ” 것과 κ°™μ•„μš”. λ‹¨μˆœνžˆ μ „νŒŒλ₯Ό 쏘고 되돌렀 λ°›λŠ” 것이 μ•„λ‹ˆλΌ 무슨 일이 μΌμ–΄λ‚˜κ³  μžˆλŠ”μ§€λ₯Ό λ³΄μ—¬μ£ΌλŠ” λ ˆμ΄λ”κ±°λ“ μš”β€ μ •λ§λ‘œ ν…ŒμŠ¬λΌμ˜ κ³ κΈ‰ μžλ™μ°¨μ— 이 κΈ°λŠ₯을 νƒ‘μž¬ν•  κ³„νšμΈκ°€? β€œμ‹€ν—˜ν•΄λ³Ό κ°€μΉ˜κ°€ μžˆμ§€μš”. λ‚˜λŠ” μ–Έμ œλ‚˜ 물리학 μ‹€ν—˜μ˜ 증거에 μ—΄λ € μžˆλŠ” μ‚¬λžŒμ΄μ—μš”.” 머슀크의 말이닀. μ—¬λ‹΄μœΌλ‘œ λ¨ΈμŠ€ν¬λŠ” 고집도 있긴 ν•˜μ§€λ§Œ 생각보닀 맀우 μ—΄λ €μžˆλŠ” μ‚¬λžŒμ΄μ—ˆκ³ , μ‹€μ œλ‘œ 생각을 λ°”κΎΈλŠ” λͺ¨μŠ΅λ„ 자주 보인닀. 그리고 FSD의 완성이 ν•œμΈ΅ κ°€κΉŒμ›Œμ‘Œλ‹€κ³  λŠκ»΄μ§„ 것이, 기쑴의 룰베이슀 방식이 μ•„λ‹Œ μ™„μ „ν•œ λ¨Έμ‹ λŸ¬λ‹ λ°©μ‹μ˜ μ˜€ν† νŒŒμΌλŸΏμ΄ μ‹€μ œλ‘œ μ›Œν‚Ήν•˜κΈ° μ‹œμž‘ν•˜λ©΄μ„œ λ¨ΈμŠ€ν¬μ— 이에 λ°˜ν•˜λŠ” λͺ¨μŠ΅μ΄ λ“±μž₯ν•œλ‹€. μ‹€μ œλ‘œ 머슀크의 2023λ…„ μ£Όμš” λͺ©ν‘œ 쀑 ν•˜λ‚˜λŠ” 도쑰λ₯Ό ν™œμš©ν•΄μ„œ AI μ‹œμŠ€ν…œμ„ ν›ˆλ ¨μ‹œν‚€λŠ” 것이기도 ν•˜κ³ . 이 λΆ€λΆ„μ—μ„œ ꡉμž₯히 μ€‘μš”ν•œ λ‚΄μš©μ΄ λ“±μž₯ν•˜λŠ”λ°, λ‰΄λŸ΄ λ„€νŠΈμ›Œν¬κ°€ 150만개의 λΉ„λ””μ˜€ 클립을 ν•™μŠ΅μ‹œν‚€λ‹ˆκΉŒ μ œλŒ€λ‘œ μž‘λ™ν•˜κΈ° μ‹œμž‘ν–ˆλ‹€λŠ” 사싀이 λ“±μž₯ν•œλ‹€. μ΄μ •λ„λ‘œ 데이터λ₯Ό λͺ¨μœΌκ³  ν•™μŠ΅μ‹œν‚¬ 수 μžˆλŠ” νšŒμ‚¬λŠ” μ „μ„Έκ³„μ—μ„œ (μ•„λ§ˆ) ν…ŒμŠ¬λΌλ°–μ— μ‘΄μž¬ν•˜κΈ° μ•ŠκΈ° λ•Œλ¬Έμ— μ—„μ²­λ‚œ 기회λ₯Ό λ§žμ΄ν•œκ²Œ λΆ„λͺ…해보인닀. 이에 λŒ€ν•΄ λ¨ΈμŠ€ν¬λŠ” β€œμš°λ¦¬λŠ” 독보적인 μœ„μΉ˜μ—μ„œ 이 일을 μˆ˜ν–‰ν•  수 μžˆμŠ΅λ‹ˆλ‹€.”라고 νšŒμ˜μ—μ„œ λ§ν–ˆλ‹€. μ˜΅ν‹°λ¨ΈμŠ€ λ‘œλ΄‡μ— λŒ€ν•œ 이야기도 μ–ΈκΈ‰λœλ‹€. λ¨ΈμŠ€ν¬κ°€ μ‚¬λžŒ ν˜•νƒœμ˜ λ‘œλ΄‡μ„ μ£Όμž₯ν•˜λŠ” 이유 λ˜ν•œ 맀우 μ§κ΄€μ μ΄μ—ˆλŠ”λ°, λŒ€λΆ€λΆ„μ˜ μž‘μ—… 곡간과 도ꡬ듀이 μ‚¬λžŒμ˜ μž‘μ—… 방식에 λ§žμΆ°μ„œ μ„€κ³„λ˜μ—ˆκΈ° λ•Œλ¬Έμ— κ·ΈλŠ” λ‘œλ΄‡ λ˜ν•œ μ‚¬λžŒμ˜ ν˜•νƒœμ— κ°€κΉŒμ›Œμ•Ό ν•œλ‹€κ³  λ―Ώκ³  μžˆλ”λΌ. λ˜ν•œ FSD에 ν™œμš©λ  도쑰λ₯Ό μ΄μš©ν•΄ ν•™μŠ΅λ˜κ³  μžˆλŠ” AIλŠ” λ‹Ήμ—°νžˆ λ‘œλ΄‡μ—λ„ μ μš©λœλ‹€λŠ”κ±Έ 확인. μ—¬λ‹΄μœΌλ‘œ λ‚˜λŠ” μ „μ„Έκ³„μ˜ λ§Žμ€ ν•˜λ“œμ›¨μ–΄λ“€ μœ„μ— ν…ŒμŠ¬λΌμ˜ λΉ„μ „ λͺ¨λΈμ΄ νƒ‘μž¬λ  수 μžˆλ‹€κ³  보고 μžˆλ‹€. 무엇보닀도 κΈ°λŒ€λ˜λŠ” 뢀뢄은 λ°”λ‘œ 이 λ¬Έμž₯. β€œκ·ΈλŠ” λ‚΄κ²Œ ν…ŒμŠ¬λΌκ°€ λ§€λ…„ 1μ‘° λ‹¬λŸ¬ μˆ˜μ΅μ„ λ‚΄λŠ” μ„Έκ³„μ—μ„œ κ°€μž₯ κ°€μΉ˜ μžˆλŠ” νšŒμ‚¬κ°€ 될 수 μžˆλŠ” ꢀ도에 μ˜¬λΌμ„°λ‹€κ³  μƒκ°ν•˜λŠ” 이유λ₯Ό μ„€λͺ…ν–ˆλ‹€.” μžμ„Έν•œ λ‚΄μš©μ€ λ‚˜μ™€μžˆμ§€ μ•Šμ§€λ§Œ, λ¨ΈμŠ€ν¬λŠ” ν…ŒμŠ¬λΌκ°€ 이미 ꢀ도에 μ˜¬λΌμ™”λ‹€κ³  μƒκ°ν•˜κ³  μžˆλ‹€. λ§ˆμ§€λ§‰μœΌλ‘œ κ°€μž₯ μ΅œκ·Όμ— μ°½μ—…ν•œ 사업체인 X.AI에 λŒ€ν•΄μ„œλ„ λ‚˜μ˜€λŠ”λ°, λ”₯λ§ˆμΈλ“œμ™€ μ˜€ν”ˆAI μΆœμ‹ μΈ 이고λ₯΄ λ°”λΆ€μŠˆν‚¨μ„ 데렀왔고 κ·Έμ—κ²Œ μ„Έ κ°€μ§€ λ―Έμ…˜μ΄ λΆ€μ—¬ν–ˆλ‹€κ³  ν•œλ‹€. 1) 컴퓨터 μ½”λ“œλ₯Ό μž‘μ„±ν•  수 μžˆλŠ” AI 봇 μ œμž‘ 2) μ •μΉ˜μ  쀑립성을 보μž₯ν•˜λŠ” μ•Œκ³ λ¦¬μ¦˜μ„ μ‚¬μš©ν•΄ 데이터 μ„ΈνŠΈλ₯Ό ν•™μŠ΅ν•˜λŠ” μ˜€ν”ˆ AI의 GPT μ‹œλ¦¬μ¦ˆμ— λŒ€ν•­ν•˜λŠ” 챗봇 μ œμž‘ 3) λ¨ΈμŠ€ν¬λŠ” β€˜μΆ”λ‘ β€™κ³Ό β€˜μ‚¬κ³ β€™λ₯Ό ν•  수 있고 β€˜μ§„λ¦¬β€™λ₯Ό κΈ°λ³Έ μ›μΉ™μœΌλ‘œ μΆ”κ΅¬ν•˜λŠ” ν˜•νƒœμ˜ μΌλ°˜μΈκ³΅μ§€λŠ₯을 λ§Œλ“œλŠ” 것. 참고둜 λ¨ΈμŠ€ν¬λŠ” OpenAIκ°€ μžμœ¨μ£Όν–‰ AIλ₯Ό λ§Œλ“œλŠ” 것보닀 본인이 LLM을 λ§Œλ“œλŠ” 것이 더 쉽닀고 λ―Ώκ³  μžˆλ‹€.

ν•΄λ‹Ή 글은 OpenAI / Y-combinator의 μƒ˜ μ•ŒνŠΈλ¨Όμ΄ 30살이 λ˜μ—ˆμ„ λ•Œ μžμ‹ μ˜ λΈ”λ‘œκ·Έμ— κ³΅κ°œν•œ κΈ€μž…λ‹ˆλ‹€. (μ›μ œ : The days are long but the decades are short) ν•΄μ™Έμ—μ„œλŠ” μ‹€λ¦¬μ½˜λ°Έλ¦¬μ˜ μ°½μ—…κ°€λ“€ μ‚¬μ΄μ—μ„œ 많이 μ½νžˆλŠ” κΈ€μΈλ°μš”. ν•œκ΅­μ—μ„œλŠ” λ²ˆμ—­λœ 적이 μ—†λŠ” 것 κ°™μ•„ 이번 κΈ°νšŒμ— 글을 λ²ˆμ—­ν•˜μ—¬ κ³΅μœ ν•©λ‹ˆλ‹€. 즐거운 ν•œκΈ€λ‚  λ˜μ‹œκΈΈ λ°”λžλ‹ˆλ‹€. https://blog.naver.com/bizucafe/223231870463

https://twitter.com/dair_ai/status/1711004647081562158 1/ LLMs Represent Space and Time - discovers that LLMs learn linear representations of space and time across multiple scales; the representations are robust to prompt variations and unified across different entity types; demonstrate that LLMs acquire fundamental structured knowledge such as space and time, claiming that language models learn beyond superficial statistics, but literal world models. https://x.com/wesg52/status/1709551516577902782?s=20