Continuous Learning_Startup & Investment
Kanalga Telegramโda oโtish
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!
Ko'proq ko'rsatish2 295
Obunachilar
-224 soatlar
-137 kunlar
-4330 kunlar
Postlar arxiv
FirstMark partner Matt Turck recalled a conversation he had with a group of engineers from Armenia, Spain and Portugal at a company offsite in Lisbon. โIt turned out every single person in that conversationโnone of whom were native English speakers or had lived in the U.S.โwas a very regular listener of the podcast,โ he said. โIt was fascinating as an example of the globalization of entrepreneurship and the Silicon Valley way of doing things.โ
https://www.theinformation.com/articles/the-besties-revenge-how-the-all-in-podcast-ate-silicon-valley?
Researchers at the Medical University of Vienna are testing a new drug matchmaking technology called Exscientia, which uses machine learning models to test the impacts of dozens of different treatments on cancer cells at the same time instead of multiple months-long courses of chemotherapy. In addition to selecting the right drug, Exscentia is leveraging machine learning to design new drugs to test on patient cells, yielding more options to sift through when searching for an effective therapy.
Expect to see more fresh food, and more big chains too. Food is an appetizing opportunity for convenience stores. On average, food-service sales accounted for about a quarter of in-store sales for convenience stores but well over a third of in-store gross profits, according to industry group NACS. At Caseyโs, the gross margin on prepared food and beverages is about 60%, while its margin on fuel is closer to 10%. And as Caseyโs Chief Executive Officer Darren Rebelez puts it: โYou might fill up a tank once a week, but you eat three times a day.โ
As drones take an increasingly prominent role in modern warfare, the need to bring them down is growing in importance.
Air defense was once the domain of specialist units with expensive equipment to destroy aircraft and missiles. Now, the proliferation of small, cheap drones is spreading the role throughout militaries, prompting governments and arms makers to work to boost their capabilities.
Pentagon acquisition chief Bill LaPlante said last week that the U.S. needed a surge in production of counterdrone technology, and that a lack of such equipment was hampering operations in both Ukraine and Israel.
โWe need lots of money, we need production lines to go up fast,โ he said.
https://wsj.com/world/this-russian-suicide-drone-is-blunting-ukraines-advance-8241a0e4
H&R Block launches AI tax filing assistant
H&R Block has launched AI Tax Assist, a new AI chatbot integrated into its paid DIY tax software, to answer tax-related questions and guide users through the filing process. This tool allows users to get detailed information about tax rules, exemptions, and specific scenarios like deductions for gig workers. H&R Block emphasizes transparency by informing users about the AI's involvement and plans to add AI-powered personalized advice. Developed with OpenAI and Microsoft, and trained exclusively on H&R Block's tax law library, AI Tax Assist aims to replicate human accountant interactions. It's available in subscription tiers starting at $35, with a focus on accuracy and user trust, backed by a team monitoring the AI for correct performance.
Intel has released their first mobile processors based on their highly anticipated Meteor Lake platform, the Core Ultra H and the Core Ultra U series. Available today, the Ultra Core H series has four SKUs, including two Ultra 7 16 core (6P+8E+2LP) chips and two 14 core (4P+8E+2LP) Ultra 5 chips. All run at a base TDP of 28 W, with a maximum turbo TDP of up to 115 W. The Core Ultra-H series is designed for ultra-portable notebooks but offers more performance in both computing and graphics within a slimline package.
Also announced is the Intel Core Ultra U-series, which includes four 15/57 W (base/turbo) SKUs, with two Core Ultra 7 and two Core Ultra 5 SKUs, and all coming with a variance in P, E-core and Intel's latest integrated Arc Xe graphics frequencies. All of Intel's announced Core Ultra U-series processors for mobile feature 10 CPU cores, with two Performance cores and eight Efficiency cores, making them ideal for lower-powered and ultra-thin notebooks.
The launch of Intel's tile-based Meteor Lake SoC marks the first step in a series of power-efficient and AI-focused chips on Intel 4 for the mobile market, which is ultimately designed to cater to the growing need to utilize AI inferencing on-chip. Both the Intel Core Ultra H and U families of chips include two new Low Power Island (LP-E) cores for low intensity workloads, with two Neural Compute Engines within the Intel AI NPU designed to tackle generative AI inferencing.
In a broader story that gets inside what the last year has been like inside the company, Jon Victor broke the news today that Google is planning a new โPixieโ assistant for its Pixel devices.
Aaron Holmes wrote about the rise of SLMs versus LLMs. (Yeah, there is already a new acronym.) SLMs, or small language models, are all about running less-powerful versions of a model on, say, a mobile phone, faster and with more protection for data. Next up is certainly glasses and other wearables, as a trio of my colleagues reported this evening.
I am excited to use all these services. Many things we do will be much easier and faster with AI, and weโll be more productive. I suspect AI wonโt reorder the business landscape so much as it will entrench the services we all use even more deeply.
Upon reflection, I think thatโs why so many in tech were panicked about the near-collapse of OpenAI. Against the big tech behemoths, it has been the one insurgent that has really broken through in usage and revenue.
Still, I worry investors havenโt gotten the memo. Natasha and Stephanie broke the story of Glean, an AI work assistant that uses OpenAI tech, getting close to raising at a $2 billion valuation with only $30 million in revenue. Those are 2021 multiples!
This week I spoke to Andreessen Horowitz general partner Anish Acharya and Initialized founder Garry Tan about a two-year-old startup financial services startup theyโve backed: The Coterie, which sells software to help wealthy individuals, fund managers and institutional investors manage their investments in venture capital funds and other alternative assets.
It seems like a tough time to grow a business focused on selling software to venture investors, given that VC firms are in the middle of their worst fundraising year in the last decade. But The Coterie and its backers argue that the startup will be able to ride out the downturn and take advantage of a bigger trend. That is, more individuals want to become VC investors, or limited partners in VC fundsโand more VC fund managers need to find limited partners beyond traditional endowments and pension funds. In other words, they need wealthy individuals with cash to spare.
VC firms, particularly new fund managers, โcan no longer rely on large institutional investors,โ said The Coterie co-founder and CEO Ethan Agarwal.
If youโre questioning whether a portal for VC investors is poised to generate the kind of returns LPs hope for, I am too. After all, it seems like the kind of product a venture capitalist would bet on, forgetting that millionaires make up less than 10% of the U.S. adult population. Whatโs more, venture investing is contracting. The number of new U.S. venture funds raised in the first three quarters of this year fell to 344, compared to over 1,400 in all of 2021, according to data firm PitchBook.
But Acharya and Tanโwho is also Y Combinatorโs CEOโsay thereโs still plenty of demand.
โIf you look at the number of people who are wealthy in Silicon Valley in tech communities, all those people have an interest in investing and being part of that community,โ said Acharya, who led a $33 million Series A investment in the company at a valuation over $100 million last year. โUsually that happens with investments in funds.โ
Other startups, particularly Carta and AngelList, have also tried to capitalize on rising demand from individuals who want shares in private assets. The Coterie doesnโt make the connection between individuals and VC investments, as AngelList does, but handles what happens once an individual or firm agrees to invest. For $100 per year for each fund an LP backs, users can wire money using The Coterieโs payment systems, review fund subscription documents and track the performance of their fundsโ holdings.
โYouโve got Robinhood to manage your public markets investments; you donโt have that for your privates,โ said Acharya, a former Credit Karma executive and an investor in $12 billionโvaluation payroll platform Deel. Tan added: โThere are a lot of people who are accredited [high-net-worth] investors who can or should have much more exposure to alternatives.โ
Perhaps thatโs the case, though many banks with large wealth operations are happy to facilitate those investments. For now the startup may be benefiting from timing: Had the company gotten started during the peak of the bull market, Agarwal said, it would likely be in a difficult situation now. But The Coterie launched in late 2021, just as the market began to turn south, which encouraged the founders to stay lean. It only has 16 employees. (Agarwal declined to disclose revenue.)
The interesting thing about this NYT/OpenAI lawsuit is the counterfactual.
If Apple is, indeed, writing substantial checks to media companies to license their content for training models, the impact of this and other lawsuits against AI companies training on non-public data will be swift and meaningful.
A very clever move by Apple if this lawsuit goes the way of NYT.
Chamath
์ ๋ฝ ๋ฏธ์ ์ฌ์์ 17์ธ๊ธฐ์ ๋ง์ด ๊ทธ๋ ค์ง ์ฅ๋ฉด์ด ์๋ค. ๋ ๋ช
์ ์ฌ์ฑ์ด ํ ๋จ์๋ฅผ ๋ถ์ก๊ณ ๋ชฉ์ ๋ฒ ๋ ๊ทธ๋ฆผ. ์๋ง ํ ๋ฒ์ฏค์ ๋ณธ ์ ์๊ฒ ์ง๋ง, ์ ๊ตฐ์ ์ฅ์๊ฐ ์ฌ๊ณ ์์ ๋ ์นผ์ ๋ค๊ณ ๋ค์ด๊ฐ ์ฃฝ์ด๋ ์ด ๊ทธ๋ฆผ์ ๊ธฐ๋
๊ต ์ฑ๊ฒฝ(์ธ๊ฒฝ)์ ๋ฑ์ฅํ๋ ์ด์ผ๊ธฐ๋ฅผ ๋ฐํ์ผ๋ก ํ๋ค.
๊ทธ๋ฐ๋ฐ ๊ฐ์ ์ฃผ์ ๋ฅผ ๊ทธ๋ฆฐ ๋ง์ ๊ทธ๋ฆผ ์ค ํ ํ๊ฐ์ ๊ทธ๋ฆผ์ด ์ ๋
๊ทธ ์ฅ๋ฉด์ ์์ํ๊ฒ ๊ทธ๋ ค์ ์ ๋ช
ํ๋ค. ์ฌ๋๋ค์ ์ ์ด ํ๊ฐ์ ๋ฒ์ ์ ์ข์ํ ๊น? ์ฌ์ฑ์ด ๊ทธ๋ ธ๊ธฐ ๋๋ฌธ์ด๋ค. 17์ธ๊ธฐ์ ํ๋ํ ์ฌ์ฑ ํ๊ฐ๋ ๊ทนํ ๋๋ฌผ๊ณ , ์ฑ๊ณต์ ์ผ๋ก ํ๋ํ ์ฌ์ฑ ํ๊ฐ๋ ๊ฑฐ์ ์ฐพ๊ธฐ ํ๋ค๋ค. ๊ทธ๋ฐ๋ฐ ์ด ํ๊ฐ, ์๋ฅดํ
๋ฏธ์์ ์ ํธ๋ ์คํค๋ 18์ธ์ ์๋ฒ์ง์ ์น๊ตฌ์๊ฒ ์ฑํญํ๋นํ๋ค.
์๋ฅดํ
๋ฏธ์์์ ์ํ๋ค์ ์ ํ์ด ์ง๋๋ฉด์ ์ฌํ์ ๊ฑฐ์ ์ํ์ก๋ค๊ฐ 20์ธ๊ธฐ์ ์๋กญ๊ฒ ์กฐ๋ช
๋๋ฉด์ ํฐ ์ธ๊ธฐ๋ฅผ ๋๊ณ ์๋ค. ํนํ 1970๋
๋ ์ดํ ๋ค์ ๋ถ๋ถ์ ํ๋ฏธ๋์ฆ์ ํ์ ์ ํธ๋ ์คํค์ ์ํ์ ๋ค์ ๋ณด๊ฒ ํ๋ ์๋๋ ฅ์ด ๋ ๊ฒ ์ฌ์ค์ด๋ค. ๊ทธ ๋ฐ๋์ ์ด ๊ทธ๋ฆผ์ ์ฑํญํ๋นํ ์ฌ์ฑ ํ๊ฐ๊ฐ ๊ทธ๋ ธ๊ธฐ ๋๋ฌธ์ ๋จ์ฑ์ ๋ํ ๋ถ๋
ธ๊ฐ ๋ด๊ฒจ์๋ ๊ฒ์ผ๋ก ํด์๋๊ณค ํ๋ค.
ํ์ง๋ง ๊ทธ๋ ๊ฒ ๋ณด๋ ๊ฒ ๋ง์๊น? ๊ทธ๋ด ์๋ ์๋ค. ํ์ง๋ง ๊ทธ๋ฐ ๋ ์ฆ๋ก ์ด ํ๊ฐ๋ฅผ ๋ฐ๋ผ๋ณด๋ฉด ๋๋ฌด๋ ๋ง์ ๊ฒ์ ๋์น๊ฒ ๋ ๋ฟ ์๋๋ผ, ์ธ๋ฌผ ์์ฒด๋ฅผ ์คํดํ๋ ๊ฒฐ๊ณผ๋ฅผ ๋ณ์ ์ ์๋ค.
ํ๊ฐ์ ์ธ์์ ์๋ ๊ฒ์ ๊ทธ์ ์ํ์ ์ดํดํ๋ ๋ฐ ๋์์ด ๋๋ค. ํ์ง๋ง ์ด์คํ ์ดํด๋ ์คํ๋ ค ์ฅ์ ๋ฌผ์ด ๋๊ณ ์๊ฐ๋ฅผ ์ด์ฐจ์์ ์ผ๋ก ๋ง๋ค ์ ์๋ค. ์ฌ๋๋ค์ ์์ ๊ฐ๊ฐ ๋จ์ฑ์ผ ๊ฒฝ์ฐ, ํนํ ๋ฐฑ์ธ ๋จ์ฑ์ผ ๊ฒฝ์ฐ ์
์ฒด์ ์ธ ์ธ๋ฌผ๋ก ์๊ฐํ๊ณ ๋ค์ํ ์๋๋ฅผ ํ์ง๋ง ์ฌ์ฑ์ด๊ฑฐ๋ ๋ฐฑ์ธ์ด ์๋๋ฉด ์๊ณ ๋ฉ์ํ ํด์์ ๊ณต๊ฐ ์์ ์ค์
๋ฃ๋๋ค.
์ค๋๋ถํฐ ์๊ฐํ๋ ๊ธ์ ๊ทธ๋ ๊ฒ ์ฝ๊ฒ ์๊ฐํ๋ ์ฌ์ฑ ์์ ๊ฐ์ ์ธ์์ ์ ๋๋ก ์ดํดํ ๊ฒฝ์ฐ ๊ทธ์ ์ํ์ ์ผ๋ง๋ ๊น๊ณ ๋ค์ธต์ ์ผ๋ก ์ดํดํ ์ ์๋์ง ๋ณด์ฌ์ฃผ๋ ์์ฃผ, ์์ฃผ ๋ฐ์ด๋ ๊ธ์ด๋ค. ์ข์ ํ๋ก ์ ๋ง๊ณ , ์ฌ์ด ํ๋ก ๋ ๋ง์ง๋ง ์ฝ๊ณ ์ข์ ํ๋ก ์ ๋๋ฌธ๋ฐ, ์ด ๊ธ์ด ๊ทธ๋ ๊ฒ ๋๋ฌผ๊ฒ ๋ณด๋ ๋ฉ์ง ๋ฏธ์ ์ฌ+ํ๋ก ์ด๋ค.
์กฐ๊ธ ๊ธธ์ง๋ง ํด์ผ์ ์ฝ๊ธฐ์ ์์ฃผ ์ข์ ๊ธ!
<์๋ฅดํ
๋ฏธ์์ ์ดํดํ๊ธฐ>
https://otterletter.com/artemisia-gentileschi/
์ค๊ตญ์์๋ Foundation Model๊ณผ AGI์ ๋ํ ๊ด์ฌ์ด ๋ง์ ๊ฒ ๊ฐ์ต๋๋ค. 12๊ฐ ํ๊ต 32๋ช
์ ์ ์๋ค์ด ํจ๊ป ์ด 169ํ์ด์ง ๋ถ๋์ Foundation Model ๊ธฐ๋ฐ ์ถ๋ก ์ ๋ํ ์ฐ๊ตฌ๋ํฅ ์กฐ์ฌ ํ๋ฆฌํ๋ฆฐํธ. ์ฌ๊ธฐ๋ 70ํ์ด์ง ์ด์์ด ์ฐธ๊ณ ๋ฌธํ ๋ชฉ๋ก.
๊ทธ๋ฆผ5์์ Foundation model ๊ธฐ๋ฐ์ Reason Task๋ค์ ๋ํ ๋ถ๋ฅ๋ฅผ ์ ์ํ๊ณ ์๋๋ฐ, 7๊ฐ ์ค๋ถ๋ฅ์ 24๊ฐ ์ธ๋ถ ๋ถ๋ฅ๋ก ๋๋์ด ๊ฐ๊ฐ์ ์ฐ๊ตฌ๋ํฅ์ ๋ํด ์๊ฐ. ์ ์ฒด ๋
ผ๋ฌธ๋ค์ ๋ค์ Github์์ ์ ๋ฆฌํด์ ์ ๊ณตํ๊ณ ์์ผ๋ ๊ด์ฌ์๋ ๋ถ๋ค์๊ฒ๋ ๋์์ด ๋์ค ๋ฏ.
์ ๋ชฉ: A Survey of Reasoning with Foundation Models: Concepts, Methodologies, and Outlook
์์ฝ:
๋ณต์กํ ๋ฌธ์ ํด๊ฒฐ์ ์ํ ์ค์ํ ๋ฅ๋ ฅ์ธ ์ถ๋ก ์ ํ์, ์๋ฃ ์ง๋จ, ๋ฒ์ฃ ์์ฌ์ ๊ฐ์ ๋ค์ํ ์ค์ ํ๊ฒฝ์์ ์ค์ถ์ ์ธ ์ญํ ์ ํฉ๋๋ค. ์ด๋ ์ธ๊ณต ์ผ๋ฐ ์ง๋ฅ(AGI) ๋ถ์ผ์ ๊ธฐ๋ณธ ๋ฐฉ๋ฒ๋ก ์ผ๋ก ์ฌ์ฉ๋ฉ๋๋ค. ๊ธฐ์ด ๋ชจ๋ธ์ด ์ง์์ ์ผ๋ก ๊ฐ๋ฐ๋จ์ ๋ฐ๋ผ ์ถ๋ก ์์
์์ ๊ทธ ๋ฅ๋ ฅ์ ํ๊ตฌํ๋ ๋ฐ ๋ํ ๊ด์ฌ์ด ๋์์ง๊ณ ์์ต๋๋ค. ์ด ๋ฐฑ์์์๋ ์ถ๋ก ์ ์ํด ์ ์๋์๊ฑฐ๋ ์ ์ฉ ๊ฐ๋ฅํ ์ค์ํ ๊ธฐ์ด ๋ชจ๋ธ์ ์๊ฐํ๊ณ ๋ค์ํ ์ถ๋ก ์์
, ๋ฐฉ๋ฒ ๋ฐ ๋ฒค์น๋งํฌ์ ์ต์ ๋ฐ์ ์ฌํญ์ ๊ฐ์กฐํฉ๋๋ค. ๊ทธ๋ฐ ๋ค์ ๊ธฐ์ด ๋ชจ๋ธ ๋ด์์ ์ถ๋ก ๋ฅ๋ ฅ์ ์ถํ ์ด๋ฉด์ ์๋ ์ ์ฌ์ ์ธ ๋ฏธ๋ ๋ฐฉํฅ์ ๋ํด ์ดํด๋ด
๋๋ค. ๋ํ ์ถ๋ก ์ ๋งฅ๋ฝ์์ ๋ฉํฐ๋ชจ๋ฌ ํ์ต, ์์จ ์์ด์ ํธ, ์ํผ ์ผ๋ผ์ธ๋จผํธ์ ๊ด๋ จ์ฑ์ ๋ํด์๋ ๋
ผ์ํฉ๋๋ค. ์ด๋ฌํ ๋ฏธ๋ ์ฐ๊ตฌ ๋ฐฉํฅ์ ๋ํด ๋
ผ์ํจ์ผ๋ก์จ ์ฐ๊ตฌ์๋ค์ด ์ด ๋ถ์ผ๋ฅผ ํ๊ตฌํ๋ ๋ฐ ์๊ฐ์ ์ป๊ณ , ๊ธฐ์ด ๋ชจ๋ธ์ ํตํ ์ถ๋ก ์ ๋ฐ์ ์ ์ด์งํ๋ฉฐ, AGI์ ๋ฐ์ ์ ๊ธฐ์ฌํ ์ ์๊ธฐ๋ฅผ ๋ฐ๋๋๋ค.
๊ฒฐ๋ก :
์ด ์กฐ์ฌ ์ฐ๊ตฌ๋ ์ถ๋ก ๋ถ์ผ์์ ๊ธฐ์ด ๋ชจ๋ธ์ ์งํ ๊ฒฝ๋ก๋ฅผ ์กฐ๋ช
ํ๋ฉฐ, ์ด๊ธฐ ๋จ๊ณ๋ถํฐ ํ์ฌ์ ๋ฐ์ ๊น์ง ๋ณต์ก์ฑ๊ณผ ํจ์จ์ฑ์ด ๋์ ๋๊ฒ ๋ฐ์ ํ์์ ๋ณด์ฌ์ค๋๋ค. ๋ฐ์ดํฐ ๊ธฐ๋ฐ ์ฌ๊ณ ์ ๊ด๋ชฉํ ๋งํ ๋ฐ์ ์ ์ธ์ ํ์ง๋ง, ๋๊ท๋ชจ ๋ชจ๋ธ์ ๊ฐ์ ๊ณผ ํ๊ณ๋ฅผ ๊ฐ๊ด์ ์ผ๋ก ์ธ์ํ๋ ๊ฒ์ด ์ค์ํฉ๋๋ค. ์ด๋ฌํ ๋งฅ๋ฝ์์ ํด์ ๊ฐ๋ฅ์ฑ๊ณผ ๋ณด์์ ๊ฐํํ๋ ๊ฒ์ ์ค์์ฑ์ ๊ฐ์กฐํ๋ ๊ฒ์ ํ์์ ์
๋๋ค. ๋ํ ์ด ์ฐ๊ตฌ์์ ์กฐ์ฌ๋ ๋ชจ๋ ๋
ผ๋ฌธ์์ ๊ธฐ์ด ๋ชจ๋ธ์ ์ถ๋ก ๋ฅ๋ ฅ์ ์ง์์ ์ผ๋ก ์ด์ธ์ ์ธ ์์ค(์๋ฅผ ๋ค์ด IMO ๋ฉ๋ฌ์ ํ๋ํ๊ฑฐ๋ ๊ณต๊ฐ ์ํ ๋ฌธ์ ๋ฅผ ํด๊ฒฐํ ์ ์๋ ์์ค)์ผ๋ก ๋์ด์ฌ๋ฆฌ๋ ๋ฐฉ๋ฒ์ ๋ํ ํฉ์๊ฐ ์์ง ์ด๋ฃจ์ด์ง์ง ์์๋ค๋ ์ ์ ์ฃผ๋ชฉํฉ๋๋ค.
๊ฒฐ๋ก ์ ์ผ๋ก, ๊ธฐ์ด ๋ชจ๋ธ์ ์ถ๋ก ์์
์์ ํฅ๋ฏธ๋ก์ด ๊ฐ๋ฅ์ฑ์ ์ ๊ณตํ์ง๋ง, ๊ทธ ๊ฐ๋ฐ๊ณผ ์ ์ฉ์ ์์ด์๋ ๋นํ์ ์ธ ์๊ฐ์ผ๋ก ์ ๊ทผํ๋ ๊ฒ์ด ํ์์ ์
๋๋ค. LLM ๊ธฐ๋ฐ ์ถ๋ก ๊ณผ ๊ด๋ จ๋ ๋์ ๊ณผ์ , ํ๊ณ, ์ํ์ ์ธ์ํ๋ ๊ฒ์ด ์ค์ํฉ๋๋ค. ์ด๋ฅผ ํตํด ์ด ๋ถ์ผ์ ์ฑ
์๊ฐ ์๊ณ ์ฌ๋ ค ๊น์ ๋ฐ์ ์ ์ด์งํ์ฌ ๊ฐ๋ ฅํ๊ณ ์ ๋ขฐํ ์ ์๋ ์ถ๋ก ์์คํ
์ ๊ฐ๋ฐ์ ๋ณด์ฅํ ์ ์์ต๋๋ค.
arXiv: https://arxiv.org/abs/2312.11562
Browse: https://browse.arxiv.org/pdf/2312.11562.pdf
PDF: https://arxiv.org/pdf/2312.11562.pdf
arXiv-vanity: https://www.arxiv-vanity.com/papers/2312.11562
Paper page: https://huggingface.co/papers/2312.11562
Papers with code: https://paperswithcode.com/paper/a-survey-of-reasoning-with-foundation-models
Github: https://github.com/reasoning-survey/awesome-reasoning-foundation-models
The Suez Canal at the northern end of the Red Sea and the Bab el-Mandeb Strait at the southern end are particularly important, with 40% of Asia-Europe trade normally passing through this area.
The U.S. has responded by forming a security initiative to protect ships in the Red Sea, involving multiple countries. Despite this, the Houthi rebels have continued their attacks, targeting at least 44 countries connected to the vessels. Shipping companies such as BP, MSC, Hapag-Lloyd, CMA CGM, and Maersk have suspended travel through the Red Sea due to the security risks.
The disruption has led to a significant increase in shipping costs and rerouting of vessels around Africa via the Cape of Good Hope, adding 7-10 days to transit times. This crisis has stirred memories of the 2021 Suez Canal blockage, which caused global trade bottlenecks
7. The current situation is causing congestion problems and could potentially spark a hike in shipping costs if it becomes an extended crisis
7.
The attacks have also raised concerns about the safety of seafarers and the potential for a wider conflict, with the Houthi leader threatening to escalate attacks to include U.S. naval ships 10. The shipping industry remains in the dark about the effectiveness of the U.S.-led Red Sea navy force and is waiting to see concrete action
11.
Maersk has indicated a return to directing its vessels through the Suez Canal, while other companies like Hapag-Lloyd and MSC continue to reroute their ships. The attacks have disrupted global shipping at a time when the Panama Canal is also less accessible due to a drought.
PandasAI is a Python library that adds Generative AI capabilities to pandas, the popular data analysis and manipulation tool. It is designed to be used in conjunction with pandas, and is not a replacement for it.
https://github.com/gventuri/pandas-ai
Googleโs GitHub Copilot competitor is now generally available and will soon use the Gemini model
Google has launched Duet AI for Developers, an AI-assisted coding tool now generally available and soon to be enhanced with the Gemini model. This platform, competing with GitHub's Copilot, offers advanced code completion and generation capabilities, bolstered by partnerships with 25 companies for specialized data and documentation. Designed to streamline coding tasks and boost productivity, Duet AI supports over 20 languages and includes features like AI log summarization and error explanation. Free until January 2024, it will subsequently cost $19 per user per month on an annual commitment.
https://techcrunch.com/2023/12/13/duet-ai-for-developers-googles-github-copilot-competitor-is-now-generally-available-and-will-soon-use-the-gemini-model/
Google unveils MedLM, a family of healthcare-focused generative AI models
Google has introduced MedLM, a new set of AI models designed for healthcare, building on its Med-PaLM 2 model. MedLM, available to certain Google Cloud customers, includes two versions for different tasks and has been trialed by healthcare organizations. This initiative is part of a larger trend among tech giants to capitalize on the healthcare AI market. However, the use of AI in healthcare has been controversial, with concerns about accuracy and privacy. Despite this, Google emphasizes its commitment to responsible AI use in healthcare.
https://techcrunch.com/2023/12/13/google-unveils-medlm-a-family-of-healthcare-focused-generative-ai-models/
1. "(์์ฆ ๊ฐ์ ์๋๋ฅผ) ์ด์๊ฐ๋ ๋ฐ ์์ด ๊ฐ์ฅ ์ค์ํ ์์ง์ ์ฌ๋๋ค์ (์ธ๋ฐ์๋) ๋ง์ ํ๋๋ฆฌ์ง ์๋ ๊ฒ๋๋ค"
2. "์ฌ๋๋ค์ ๋ค๋ค ์์ ๋ง์ ์๊ฐ์ด ์์ ๊ฒ์ด๊ณ , ์์ ์ ๋ง์ ๋ค์ง ์์ง ์์ผ๋ฉด ํญ์ ๊ถ์๋ ๋์ฃ . ๊ทธ๋ด ๋์ผ์๋ก ์ฐ๋ฆฌ๋ ๊ฐํด์ ธ์ผ ํฉ๋๋ค"
3. ๊ทธ๋ด ๋์ผ์๋ก ๋
๋ฆฝ์ ์ผ๋ก ์๊ฐํ๊ณ , ์์ ์ ์ ๋
์ ์งํค๊ธฐ ์ํด ์ธ์์ผ ํฉ๋๋ค. (์ฝ๊ฒ ๋จ์) ๋น๋ํ๋ ๊ฑธ ์ฆ๊ธฐ๋ ์ฌ๋๋ค๋ก๋ถํฐ ๋ง์ด์ฃ "
4. "์ ๋ ์ด๊ฒ ์ง๊ธ ์๋์ ๊ฐ์ฅ ํ์ํ ๋ฅ๋ ฅ์ด๋ผ๊ณ ์๊ฐํฉ๋๋ค"
- ์กฐ์ ๋ฌด๋ฆฌ๋ด
ํด์ฌ ์ดํ, ๋ถ์ง๋ฐํ ์ฌ๋๋ค์ ๋ง๋๋ฉด์ ๋ด๊ฐ ๋ง๋๋ ์ด๋ค์ ํน์ง์ ํ๋ ์๊ฒ ๋๋ค. ๊ทธ๊ฒ์ ๋ด๊ฐ ๋ง๋๊ณ ์ถ์ดํ๋ ์ฌ๋๋ค์ ๋ค๋ค ์๊ธฐ๋ง์ '์ธ๊ณ'๋ ๊ฒ์ ๊ฐ๊ณ ์๋ ์ฌ๋๋ค์ด๋ผ๋ ์ ์ด๋ค. ๋ฌ๋ฆฌ ๋งํ๋ฉด, ์๊ธฐ๋ง์ 'ํ์ด ์๋ ๋ง์'์ด ์์ด์, ๊ทธ ๋ง์์ ์ค์ฌ์ผ๋ก ์๊ธฐ ์ถ์ ๋ณดํธ๋ง์ ์น ์ค ์๋ ์ฌ๋๋ค์ด๋ผ๊ณ ๋ ํ ์ ์๋ค.
์๊ธฐ ๋ง์์ ํ์ด ์๋ ์ฌ๋์, ๊ทธ๋ฅ ํ์ธ๋ค์๊ฒ ํฉ์ธ๋ ค ์ฐ๋ค. ๋จ๋ค์ด ์ข๋ค๋ ๊ฒ๋ง์ ๋ฐ๋ผ๊ฐ๊ธฐ ๋ฐ์๊ณ , ๋จ๋ค๊ณผ์ ์ฐ์ด์ ๋๋๋ ๋น๊ต์ ๋ช์์ ๋น ์ ธ๋์ฌ ๋ฐฉ๋ฒ์ ๋ชจ๋ฅธ๋ค. ํญ์ ํ์ธ๋ค์ ์ ๊ฒฝ ์ฐ๋ฉด์, ๋จ๋ค์ด ์์ ์ ์ฐ์ต๊ฒ ๋ณด์ง๋ ์์์ง, ๋ด๊ฐ ๋จ๋ค๋ณด๋ค ์๋๊ฑฐ๋ ๋ชป๋์ง๋ ์์์ง ์ ๊ฒฝ์ฐ๋ฉฐ, ํ์ธ๊ณผ '๊ฐ์ ๋ ์ด์ค'์์ ์๋ก๋ฅผ ๋น๊ตํ๊ธฐ ๋ฐ์๋ค. ๊ทธ๋ค์๊ฒ ๋ฒค์ธ ๊ฐ ์๋ ์๋ํ๋ฅผ ํ๋ ์ถ, ์๋ฅด๋ฉ์ค๊ฐ ์๋ ์ค๋ฌ์ ๋งค๋ ๊ฒ์ ๋ฌด์กฐ๊ฑด ์ ์๋ณด๋ค ํ์๊ฐ ์ด๋ฑํ ๊ฒ์ด๋ค.
๋ง์ฐฌ๊ฐ์ง๋ก ํ์ฌ๊ฐ ์๋ ๋ณํธ์ฌ์ ์ถ, ๊ฐ๋จ์ด ์๋ ๊ฐ๋ถ์ ์ถ, ๊ณจํ ์น๋ ์ถ์ด ์๋ ๋๋ค ๊ณต์์ ๋ฌ๋ฆฌ๋ ์ถ์ ์ญ์ ๋ ์ด๋ฑํ ๊ฒ์ด๋ค. ํฌ๋ฆฌ์ค๋ง์ค ์ฐํด์ 100๋ง ์์ง๋ฆฌ ํธํ
์ ์ก๊ณ ์ค๋ง์นด์ธ๋ฅผ ์ฆ๊ธฐ๋ ์ถ์, ์ง์์ ๊ฐ์กฑ๋ผ๋ฆฌ ๋ฉ๋ก ์ ๊น์๋จน๊ณ ์์ ํธ๋ฆฌ๋ฅผ ๊พธ๋ฏธ๊ณ ๋
ธ๋ ์ถ๋ณด๋ค '์ด๊ฒฌ์ ์ฌ์ง ์์ด' ์ฐ์ํ ๊ฒ์ด๋ค. ๊ฐ์ธ์ ์ผ๋ก ๋๋ ๊ทธ๋ฐ ์ฌ๊ณ ๊ด์ด ๋จธ๋ฆฟ์์ ๋๋ฌด ๊น์ด ๋ฐํ๋ฒ๋ฆฐ ์ฌ๋๋ค๊ณผ๋ ํ ์๊ฐ ์ด์ ๋ง์ฃผ์์ ์จ์ฌ๊ธฐ๊ฐ ์ด๋ ต๋ค. ๊ทธ ์์๋ ฅ์ ๋ถ์ฌ, ๋ง์์ ๋น๊ณคํจ, ํ์ผํ๋์ด ๊ฑฐ์ ๊ธฐ๊ณ๊ฐ ๋์ด๋ฒ๋ฆฐ ๋น๊ต์ ํผํด์์์ ๋ง์์ด ๊ฒฌ๋๊ธฐ ์ด๋ ต๋ค.
๋ด๊ฐ ํด์ฌ ์ดํ์ ์ฐพ์๋ค๋๋ฉฐ ์ฝ์์ก๊ณ ๋ง๋๋ฉฐ ๋์ด์๊ฐ์ฉ, ํน์ ๊ทธ ์ด์์ฉ ๋ง์ฃผ์์ ์ด์ผ๊ธฐ ๋๋๋ ์ฌ๋๋ค ์ค์๋ ๊ทธ๋ฐ ์ฌ๋์ด ํ๋๋ ์์๋ค๋ ๊ฑธ ๊นจ๋ฌ์๋ค. ๊ทธ๋ค์๊ฒ๋ ๋ชจ๋ ์ ๋ง๋ค์ ์ง์งํ ์ธ๊ณ์ ๋ง์์ด๋ ๊ฒ ์์ด์, ๊ทธ ๋ฐ๊นฅ์ ํ์ค์ด๋ ๋น๊ต์์์ ๋ ์ด์ค๋ก ์ ์ฒ ๋ 'ํ์์ ์ธ๊ณ'๊ฐ ์นจ๋ฒํ์ง ๋ชปํ๋ ๋ฒฝ์ด ์๋ค. ์ด๋ฅผํ
๋ฉด, ๊ทธ๋ค์ ํผ๊ณคํ ๋ชจ๋ ํผํด์์์ ์ธ ํ์ค ๋ณด๋ค๋ ์ฑ
ํ ๊ถ ์ฝ๋ ๋ฐค, ์ฐ ์์ ๊ฑฐ๋๋ ์ผ, ์๊ธฐ๋ง์ ์ธ๊ณ๋ฅผ ์ฐฝ์ํ๋ ์ผ์ ๋ชฐ๋ํ๋ ๊ฑธ ๋ ์ข์ํ๋ค.
์ธ์ ๊ฐ ๋๋, ์ธ์์ ์๋ง์ ๋ฌธ์ ๋ค์ด ๊ฒฐ๊ตญ '์ฑ
ํ ๊ถ ์ฝ๋ ๋ฐค'์ ์ฌ๋ํ ์ค ๋ชจ๋ฅด๋ ๋ง์์์ ์ค๋์ง๋ ๋ชจ๋ฅธ๋ค๊ณ ์ด ์ ์ด ์๋ค. ๋น์ ์ ์ธ ์ด์ผ๊ธฐ์์ง๋ง, ํต์ฌ์ ๋ค๋ฅด์ง ์๋ค๊ณ ์๊ฐํ๋ค. ์ฌ์ค, ๊ทธ๋ ๊ฒ ํ์์ค๋ฝ๊ฒ ๋๊ณผ ๊ถ๋ ฅ์ ๊ฐ๊ตฌํ๋ ์ฌ๋๋ค์ด ๊ฒฐ๊ตญ ๋ฌด์์ ํ๋์ง ๋ณด๋ฉด, ๊ทธ๋ฅ ๋๋จํ ๊ฑธ ํ๋ ๊ฒ๋ ์๋๋ค. ๊ธฐ๊ปํด์ผ ๋ชฐ๋ ๋ฃธ์๋กฑ์์ ์ ๋จน๊ณ ์ ํฅ์
์์์ ๋ง์ฝํ๋ค๊ฐ ๊ฑธ๋ ค์ ์ดํผ๋นํ๊ฑฐ๋, ๋ถํ์ง์ ์ฑ์ถํํ๊ณ , ๊ณจํ ๋ช ๋ฒ ์น๋ฌ ๋ค๋๋ ๊ฒ ์ธ์ ๊ทธ๋ฆฌ ๋๋จํ ๊ฑธ ํ๋ค๋ ์๊ธธ ๋ค์ด๋ณธ ์ ์ด ์๋ค. ๋ง์ฝ ๊ทธ๊ฐ ๊ทธ๋ฅ ์ฑ
ํ ๊ถ ์ฝ๋ ๊ฑธ ๋ฌด์๋ณด๋ค ์ฌ๋ํ ์ค ์๋ ๋ง์์ด ์์๋ค๋ฉด, ๊ทธ๋ ๊ฒ ์ถ์ ํ๋นํ ํ์๋ ์์์ ๊ฒ์ด๋ค.
์ฌํด ๋๋ ์ธ์์ผ๋ก๋ถํฐ, ๊ทธ ์ธ์์ ์ฌ๋ฌ ์ฌ๋๋ค๋ก๋ถํฐ ๋๋ฌด๋ ๋ง์ ๊ฑธ ๋ฐฐ์ ๋ค๋ ์๊ฐ์ด ๋ ๋ค. ๊ทธ๋ฌ๋ฉด์ ๋ด๊ฒ ์ข์ ์ฌ๋๋ค, ๊ทธ๋ฌ๋๊น ๋์ ๋ง๋ ์ฌ๋๋ค๊ณผ ๊ทธ๋ ์ง ์์ ์ฌ๋๋ค์ ๊ตฌ๋ณํ ์ค๋ ์๊ฒ ๋์๋ค. ๋์๊ฐ ์ถ์์ ์ค์ํ ๊ฒ์, ๋ด๊ฐ ์ธ๊ณ๋ฅผ ๊ณต์ ํ๊ณ ์ถ์ ์ฌ๋๋ค๋ง์ ๊ณ์ ๋๋ฉฐ ์ถ์ ํ์ฅํด๊ฐ๋ ์ผ์ด๋ผ๋ ๊ฒ๋ ๋ ๋ช
ํํ ๋ฐฐ์ ๋ค. ์ถ์ ์๊ธฐ๊ฐ ์ํ๋, ์ข์ ๋ฐฉ์์ผ๋ก ๋ง๋ค์ด๊ฐ๋ ๋ฐฉ๋ฒ์ ์์ธ๋ก ๊ฐ๋จํ๋ค. ๊ทธ ๋ฐฉ์์ ํด์น๋ ์ฌ๋๋ค์ ๊ฑธ๋ฌ๋ด๊ณ , ๊ทธ ๋ฐฉ์์ ๋์์ ์ฃผ๋ ์ด๋ค์ ์์ ๋ถ์ก๋ ๊ฒ์ด๋ค.
์ฌํด ๋ด๊ฒ๋ ์ฐํ๋ค์ด ์์๋ค. ๊ทธ๋ค์ ์ ๋ง๋ค์ ์ง๊ธ๋ฒจ์ ๋ค๋ ค์ฃผ์๊ณ , ๋๋ ๊ทธ ์์์ ์กฐ๊ธ ๋ ๋๋ ทํ๊ฒ ๋์ ๋ง์๊ณผ ์ถ์ ์๊ฒ ๋์๋ค. ์ํด์๋ ์กฐ๊ธ ๋ ๋ช
๋ฃํด์ง ๋ง์์ผ๋ก ์ถ์ ์ด์๊ฐ ์ ์์ ๊ฑฐ๋ผ ๋ฏฟ์ด์ง๋ค. ์ด ๋ฏฟ์ด์ง์ด ๋ฌํ๊ณ ๋ฐ๋ปํ๊ณ ๊ฐ๋ ฌํ๋ค.
์ ์ง์ฐ๋
Endi mavjud! Telegram Tadqiqoti 2025 โ yilning asosiy insaytlari 
