AI Deeply @AiDeeply
AI is reshaping the world. Who are the people and companies driving the change? Visit our website to search more than 5,000 profiles. Joined November 2022-
Tweets2K
-
Followers397
-
Following5K
-
Likes2K
Worth reading. Covers some of the key points behind potential new approach to AI / ML computing.
This: "Alignment is tricky not just because of technical reasons or statistical choices but also for messy, normative and data-centric human factors." The most important alignment question is "to what?" and therefore "who decides?".
This: "Alignment is tricky not just because of technical reasons or statistical choices but also for messy, normative and data-centric human factors." The most important alignment question is "to what?" and therefore "who decides?".
Benchmarking is hard ... but this helps: "the best eval is the one you build yourself for your own needs and show no one"
Benchmarking is hard ... but this helps: "the best eval is the one you build yourself for your own needs and show no one"
Very interesting combination of techniques. #LocalFirst AI Or at least #GPUPoor AI (Click through to the thread for link to more details.)
Very interesting combination of techniques. #LocalFirst AI Or at least #GPUPoor AI (Click through to the thread for link to more details.)
Amazing but inherently limited: "even if you train LLMs only on factual data*, LLMs …[will]… produce completions that are not factual!" [due to] "the basic n-gram structure" "*and I will suspend my disbelief … about the impossibility of doing that in a multi-polar world"
Amazing but inherently limited: "even if you train LLMs only on factual data*, LLMs …[will]… produce completions that are not factual!" [due to] "the basic n-gram structure" "*and I will suspend my disbelief … about the impossibility of doing that in a multi-polar world"
AI IDE in the cloud: Several announcements from @Replit yesterday including Code Repair (apparently beats GPT-4 and Claude 3 Opus) and Teams (coming soon). More context:
"AI could actually help rebuild the middle class". Worth highlighting per the NYT profile.
"AI could actually help rebuild the middle class". Worth highlighting per the NYT profile.
These AI integrations with tools aimed at ordinary business users may open up some real productivity improvements. (And likely some frustration with current limitations that are so far inherent to LLMs.)
These AI integrations with tools aimed at ordinary business users may open up some real productivity improvements. (And likely some frustration with current limitations that are so far inherent to LLMs.)
"The actual security mindset ... says that failures are inevitable ... Therefore ... you need to build systems to be as anti-fragile as possible, as robust as possible even in the presence of failure, incorporating multiple layers of redundant defenses..."
"The actual security mindset ... says that failures are inevitable ... Therefore ... you need to build systems to be as anti-fragile as possible, as robust as possible even in the presence of failure, incorporating multiple layers of redundant defenses..."
AI regulation gets prominent mention by independent VP candidate. CC @aftfuture and @a16z [Starting 4:01] “AI unregulated can be disastrous for us. We're already seeing early signs of it. [NYT headline: A.I. Poses 'Risk of Extinction,' Industry Leaders Warn by Kevin Roose |…
AI regulation gets prominent mention by independent VP candidate. CC @aftfuture and @a16z [Starting 4:01] “AI unregulated can be disastrous for us. We're already seeing early signs of it. [NYT headline: A.I. Poses 'Risk of Extinction,' Industry Leaders Warn by Kevin Roose |…
Generating more and better tests is a great use of AI. Will be especially important as more software gets written (in part or whole) by AI.
Generating more and better tests is a great use of AI. Will be especially important as more software gets written (in part or whole) by AI.
No surprise that Saudi Arabia is investing oil money in AI. I wonder if there's a convenient list of Arxiv papers from people at @AI_KAUST and @KaustVision and @KaustVisionCAIR
No surprise that Saudi Arabia is investing oil money in AI. I wonder if there's a convenient list of Arxiv papers from people at @AI_KAUST and @KaustVision and @KaustVisionCAIR
The cost of regulation: "The EU approved its AI Act last week, considered one of the world’s strictest regimes over the technology, causing companies that offer services to European consumers to scale up compliance teams. “This is a cost that weighs heavy on some of those…
The cost of regulation: "The EU approved its AI Act last week, considered one of the world’s strictest regimes over the technology, causing companies that offer services to European consumers to scale up compliance teams. “This is a cost that weighs heavy on some of those…
Given $1.3B raised, this part is important: "Microsoft is making the Inflection shareholders whole via a licensing deal… and they still keep their shares in Inflection, which is an ongoing concern."
Given $1.3B raised, this part is important: "Microsoft is making the Inflection shareholders whole via a licensing deal… and they still keep their shares in Inflection, which is an ongoing concern."
AI in healthcare is both important and risky. Additional funding is good.
AI in healthcare is both important and risky. Additional funding is good.
Bold claims. "state of the art in structured tasks such as extraction and function calling while running up to 100× faster, with 10× lower latency, outputting 100% reliable structure" see also: github.com/rysana-ai/rysa… MIT license; no recent updates
Bold claims. "state of the art in structured tasks such as extraction and function calling while running up to 100× faster, with 10× lower latency, outputting 100% reliable structure" see also: github.com/rysana-ai/rysa… MIT license; no recent updates
Agreed. Claude & other LLMs are "alive" in the same sense that Sherlock Holmes is alive. With good fiction, we get caught up in the story and it feels real. Then we turn the last page of the book, and can discuss motivation and other details without calling it "alive".
Agreed. Claude & other LLMs are "alive" in the same sense that Sherlock Holmes is alive. With good fiction, we get caught up in the story and it feels real. Then we turn the last page of the book, and can discuss motivation and other details without calling it "alive".
Great example of @levie's point: "AI will be taking functions or processes in business that are artificially constrained by their cost or complexity, and scaling them up to 100X more customers"
Great example of @levie's point: "AI will be taking functions or processes in business that are artificially constrained by their cost or complexity, and scaling them up to 100X more customers"
Mark you calendars.
Nice.
กระดิ @xsSh5tg04a95ZoH
44 Followers 1K Following ต้องการมีทักษะการออกเดทที่น่าทึ่งหรือไม่? ที่นี่มีสิ่งมหัศจรรย์ที่คุณไม่ควรพลาดฉันจะอัปเดตข้อมูลการติดต่อในหน้าแรกFrances @frances_leonhar
763 Followers 3K FollowingMia Perkins @MiaPerkins82216
131 Followers 3K FollowingOlive @olive_walters93
796 Followers 3K FollowingTheWrist @ZeWrist
1 Followers 19 FollowingWalid Saba @sabawalid
971 Followers 1K Following PhD Comp Science (AI/NLU); 25 years experience (AT&T Bell Labs, IBM, and AIR); 40+ publications including an award winning paperJames Prince @SJxPrince
194 Followers 3K Following Multimodal Autonomous Agents @ the point of customer acquisition / Inbound + outbound workflows on Whatsapp, 🌐, ☎️ / https://t.co/RqVaraSMsN /கற்றது கை மண் அளவுRyan.M.Nefdt @ryan_nefdt
167 Followers 134 Following Philosopher of science in linguistics, cognitive science, & AI. Author of 'Language, Science, and Structure' @OxUniPress, PhD @StAndrewsPhil, work @UCT_ResearchJohn Cena @JohnCena
14.3M Followers 777K Following A forum of thoughts and perspectives designed to ignite conversations and actions leading to growth, and occasional self promotion. #NeverGiveUp #RiseAboveHateมะลวิภา @STCzWqB95up4C
62 Followers 1K Following ความเซ็กซี่มีมากกว่าหนึ่งด้าน ติดตามฉันและค้นพบช่วงเวลาอื่นๆ ที่จะทำให้หัวใจคุณเต้นเร็วขึ้น! หน้าแรกของข้อมูลการติดต่อจะได้รับการอัปเดตตลอดเวลาBaya Systems @bayasystems
1K Followers 3K Following Advancing the semiconductor industry with groundbreaking innovations in fabric IP and software solutionsSimon McCoy @SimonMcTLOSTL
1K Followers 2K Following My name is Simon McCoy, I am the author of the 'To Live Outside The Law' series of books that recount my real life adventures as an outlaw.Mingchen Zhuge @MingchenZhuge
119 Followers 116 Following PhD Student in @AI_KAUST; Fortune to have @SchmidhuberAI as my advisor; Contributing to @MetaGPT_ and https://t.co/wnoau7Zhp6Matheus @mathdesilva
126 Followers 357 FollowingCiprian Cîmpan @devnulli
355 Followers 2K Following I build your next-gen products in my resilient AI cloud @fifi_ai 🤖 Full-stack engineer obsessed with DevOps 🚀Vladimir @v4fs_
62 Followers 236 FollowingCatherine @careycatherine6
218 Followers 3K FollowingShawn Chauhan @shawnchauhan1
4K Followers 1K Following The AI Guy | Diving deep into AI & tech for you | Creator of https://t.co/zgmUTDIwzP | Interested in teaming up? DM me.Jennifer Lanford @JenniferLa6495
8 Followers 725 FollowingInari @useinari
107 Followers 394 Following An AI-powered product discovery and feedback analytics tool. Surface insights and product opportunities from your customer data auto-magically using AI.Bhanu Teja Adem @AdemTeja
3 Followers 51 FollowingSolitude @SolitudeAgents
34 Followers 109 Following A marketplace to discover AI Agents that detect and automate complex workflows for you.Joe Zimmerman @JoeZimmerm67335
129 Followers 3K FollowingAhmed Morsi @eramax
26 Followers 631 Followingeba2cq650uk @njd738fl8
17 Followers 963 Following We first transfer USDT to you TRC20, you return 90% to BEP20, you get 10% , 2K per day Our co hv a large amt of USDT need to from TRC20 convert to BEP20 networkCarolyn Jensen @CarolynJ63498
114 Followers 3K FollowingNikhil Delacruz @DelacruzNi79639
11 Followers 655 FollowingTino @PriNova75
37 Followers 93 Following AI Enthusiast, Developer since over 20 years, CG creator, tutor and digital nomad discovering nature.Ho Leung Ng (吴浩�.. @AI_drugs
2K Followers 5K Following AI/computational drug discovery, seeking new opportunitiesJack Reacher @JackReach516
71 Followers 1K FollowingAshutosh Mehra @ashutoshmehra
1K Followers 5K Following Senior Principal Scientist at Adobe. Working on Acrobat AI Assistant, LLMs, and document ML.PI Health Cancer Hosp.. @pihealthcancer
18 Followers 27 Following Revolutionary Care, Transformative Research Hyderabad, India 𝐂𝐨𝐧𝐭𝐚𝐜𝐭 𝐮𝐬 𝐚𝐭 𝟎𝟒𝟎 𝟔𝟗𝟒𝟓 𝟖𝟐𝟎𝟏 #pihealthcancerhospital #cancerAaditya ; @Aaditya26082004
531 Followers 7K Following CS'26 • Machine Learning • Open-Source • Web Dev. • Algorithms • Jai Shree Krishna 🦚🪈Alpay Ariyak @AlpayAriyak
1K Followers 2K Following AI @RunPod_io | Lead: @OpenChatDev (600k+ downloads on HuggingFace🤗)jona @_jon_gg
98 Followers 632 FollowingLisa Davis @LisaDavis397665
13 Followers 692 FollowingGab e/acc @gluisvieira
40 Followers 1K FollowingKaren Clark @clark_kare71964
14 Followers 731 FollowingNan HUO @NanHUO9637
209 Followers 420 Following CS PhD Student @HKUniversity. Previously M.S. @JohnsHopkins.louis030195 @louis030195
1K Followers 2K Following ai defense founder | ex 🕵️ | top secret clearance | leukemia survivor @ 13 | 🥊 muay thai | @techstars | @OrangeDAOxyz | 🤖 LLMs since gpt-1ManivMani ManivMani @ManivmaniM
12 Followers 282 FollowingYixin Wan @yixin_wan_
1K Followers 847 Following PhD student @UCLAComSci | Trustworthy Generative Models | Previously @AmazonScience, @MSFTResearch AsiaIzzy @areweoutside
6 Followers 10 Followingfifi.ai @fifi_ai
4 Followers 6 Following Run Advanced AI, Worry-Free. We've Got the Infrastructure Covered.Mindcorp.ai @mindcorpai
58 Followers 41 Following https://t.co/t1OylwMmhI Mindcorp AI develops cognitive agents for knowledge work on our proprietary Cognition platform.Nova Spivack @novaspivack
35K Followers 728 Following Innovator / entrepreneur / futurist / venture producer. AI, AR, space, pharma, analytics, science, finance. Full Bio: https://t.co/JqBSOz8pEgLouis Anslow @LouisAnslow
3K Followers 5K Following Curator of @PessimistsArc • Rethinking messages @TellChat • https://t.co/30JgUOkiBD • AI Doomer Beat Reporter • Email: [email protected]Mukesh Jha @jmukesh99
168 Followers 854 Following Engineer × Entrepreneur 🗿 NLP • LLM Agent • DAIICT'24 👨🎓 Building AI products 🤖👨💻 Talks about AI, Tech, SaaS, Entrepreneurship🗽 Let's connect 👋Elliot Vaucher @ElliotVaucher
76 Followers 415 Following Passionate about intelligence, artificial or not. #llmops. My dream is to bridge the gap between Claude Shannon, B.F Skinner and C.G Jung🇨🇭 https://t.co/cbYjitRuoREmilia David @miyadavid
2K Followers 298 Following AI reporter @verge | Was @BusinessInsider @VCJournal @WatersTech @AMM1882 @bworldph | (Beware the occasional fandom tweet) | She/HerSpiros Margaris @SpirosMargaris
131K Followers 23K Following #VC | No 1 #Fintech #Banking @Refinitiv & @Onalytica | #AI | @TEDx | @qualco_sa @natechsa @SparkLabsGlobal @ai_mediastalker @investwithCARL @HeradoHQ @barraqappAirtable @airtable
60K Followers 857 Following Build powerful work apps, without coding. For help visit https://t.co/w3nNw9o6Ae; @airtablestatus for status updates; @airtabledev for platform newsRaghav Sethi @raghavsethi
762 Followers 952 Following Engineering @Airtable. Committer @trinodb. Previously @PrincetonCS, @IIITDelhi. Views are my own.Joshua Ma @__joshma
1K Followers 1K Following ai at @airtable - prev founder @airplanedev, cto @benchlingAirplane @AirplaneDev
2K Followers 2 Following Developer platform for internal tooling and workflow automation.Alex Trott @alexrtrott
699 Followers 269 Following Research @DbrxMosaicAI. Neuroscience PhD in a previous life. Whispering models into sentience one parameter at a time. (opinions are my own.)amit ⚡️ @gravicle
7K Followers 4K Following ceo @LumaLabsAI | prev: built Vision Pro at | everything is figureoutableDan Kondratyuk @hyperparticle
391 Followers 343 Following Working on Generative Models @LumaLabsAI. Prev. #VideoPoet @GoogleAI. I'm a developer that enjoys solving puzzles, one piece at a time.Alan Cowen @AlanCowen
2K Followers 206 Following CEO @hume_ai, teaching AI to make people happy. AI researcher + emotion scientist. Prev @googledeepmind⟁ndrew V @AndrewVoirol
3K Followers 5K Following GenAI | AI Research | AI Engineer. Tinkering with tenacity. Bridging bytes with biology. When life throws curveballs, I code the comeback. Building from 0 to 1.Erik Kaunismäki @ErikKaum
343 Followers 502 Following SWE @BananaDev_ 🍌 | @_buildspace s1 | https://t.co/sdWIWr6tCQAbhi Venigalla @abhi_venigalla
5K Followers 1K Following Researcher @Databricks. Former @MosaicML, @CerebrasSystems. Addicted to all things compute.Deepak Subramani @deepakns
1K Followers 595 Following Asst. Prof. Computational and Data Sciences, @iiscbangalore. PhD, @MIT. BTech, @iitmadras. AI, ML, DL, Climate, Upskilling, Education.Yucheng Li @liyucheng_2
181 Followers 209 Following NLPer at University of Surrey @UniOfSurrey, UK. Research Intern at @MSFTResearch.Sasank Chilamkurthy @sasank51
2K Followers 440 Following Working on something new at the intersection of AI and hardware.Mithril Security @MithrilSecurity
242 Followers 12 Following ⚙️ The first simple privacy framework for data science collaboration You can check our open-source projects 👉 https://t.co/pdVP9TvZs4…Daniel Huynh @dhuynh95
898 Followers 220 Following CEO at @MithrilSecurity, AI & privacy startup. Lead of 🌊LaVague, open-source Large Action Model project https://t.co/D4n9bzUjncSamuel Hammond 🌐�.. @hamandcheese
22K Followers 2K Following Senior economist @joinFAI. Nonresident fellow @NiskanenCenter. Pluralist. 'The world is second best, at best.' | [email protected]Rui @Rui45898440
173 Followers 456 Following @OptimalScale maintainer. Interested in Optimization, Acceleration, LLMTianfu Wang @TianfuWang2
25 Followers 52 FollowingHuaizu Jiang @HuaizuJiang
2K Followers 1K Following Assistant Professor at Northeastern University. Previously Postdoc at Caltech and Visiting Researcher at NVIDIA. PhD from UMass Amherst.Joshua Batson @thebasepoint
2K Followers 707 Following trying to understand evolved systems (🖥 and 🧬) interpretability research @anthropicai formerly @czbiohub, @mit mathMirascope AI @mirascopeai
93 Followers 178 Following Mirascope is a Python toolkit for working with LLMs. Building with Mirascope feels just like writing the Python code you’re already used to. Built on Pydantic 2Sreekanth Mukku @sreekanth_mukku
1K Followers 3K Following @NetworkdFutures Working on Sustainable AI Governance, Data commons and Inequalities. Prev @hippoai @CAISnrw @knnktv @IBM @Oracle @Deloitte. Alum @BrandtSchoolsearch founder @n0riskn0r3ward
376 Followers 900 Following Solo entrepreneur passionate about search tech. Self-taught dev building a niche search product and sharing what I learn along the way.Brett Larsen @_BrettLarsen
419 Followers 332 Following Sr. Research Scientist @DbrxMosaicAI | Guest Researcher @FlatironInst @NYU_CNS | Efficient deep learning + better algorithms for data scienceZack Ankner @ZackAnkner
485 Followers 305 Following Junior @MIT. President of AI@MIT. Research Scientist Intern @MosaicML. A(CL)verage Embargo enjoyer.Linden Li @lindensli
1K Followers 534 Following CS @Stanford, @StanfordSVL. Research/Eng @MosaicML, previously @NVIDIA.Tal Peretz @talperetz_
136 Followers 35 Following Building AI products @Zapier. Co-founder @timeOSai (formerly Magical)salesstack.ai @SalesStackAI
35 Followers 2 Following SalesStack is a custom Web Browser powering Ghost, the world's first AI Sales Employee that can sell on autoplay using its Full Self Browsing Capabilities.Rahmad Mahendra @rmahendrarm
74 Followers 313 Following NLP, Badminton 🇮🇩 | PhD student @ARC_AIMedTech @RMITComputing | @FASILKOM_UILintang Sutawika @lintangsutawika
381 Followers 565 Following Incoming Ph.D. student @LTIatCMU. Researcher at @AIEleuther. Maintainer of LM-Eval Harness. Here for machine learning papers and discussion.Maurice Peemen @MauricePeemen
20 Followers 14 FollowingTed Xiao @xiao_ted
11K Followers 682 Following I teach robots to be smarter @GoogleDeepMind. Tweets about robot learning, scaling, and large models. Opinions my own.(1/7) Do you want to test code generation models on the domains you care about? Struggling to find existing benchmarks that suit your needs? Our new work *CodeBenchGen* helps you build execution-based benchmarks based on your selected code fragments! (arxiv.org/abs/2404.00566)
I'm excited about Yiqing's work on a framework for creating code-generation benchmarks from naturally-occurring code, e.g. from GitHub. We use test cases to evaluate! As a proof-of-concept, we create a new benchmark from (subsets of) CodeSearchNet.
(1/7) Do you want to test code generation models on the domains you care about? Struggling to find existing benchmarks that suit your needs? Our new work *CodeBenchGen* helps you build execution-based benchmarks based on your selected code fragments! (arxiv.org/abs/2404.00566)
Replacing Judges with Juries Evaluating LLM Generations with a Panel of Diverse Models As Large Language Models (LLMs) have become more advanced, they have outpaced our abilities to accurately evaluate their quality. Not only is finding data to adequately probe
I have finally completed the tutorial on using Spacy-llm and DSPy for Named Entity Recognition. It concludes with an unexpected twist. Link in the following tweet. Preview:
We don't talk enough about spacy-llm. It is so powerful. And when you mix it with DSPy, the results are incredible. I am working on a video that i will add to the "Advanced DSPy" module on Lycee AI. It will be lit 🔥🔥🔥
LLMs as a judge has been widely accepted as a workable replacement of human eval but relying on a single model introduces systematic bias. Happy to share a new paper from our team led by @pat_verga that shows a panel of models as judge offers a more accurate and cheaper solution.
New paper from our team, led by @pat_verga Are you: * Doing evaluation with LLMs? * Using a huge model? * Worried about self-recognition? Try an ensemble of smaller LLMs. Use a PoLL: less biased, faster, 7x cheaper. Works great on QA & Arena-hard evals arxiv.org/abs/2404.18796
If Model A beats B in benchmarks, is it really better? Not if it trained on those benchmarks—that's an unfair edge! How can you tell if a model used benchmark data for training? 🤔 Welcome to check out our latest work: huggingface.co/papers/2404.18… (1/n)
Are your LLMs highly accurate, or simply contaminated? As the race to build the best LLM intensifies, clean evaluation is becoming more important than ever, yet contaminated LLMs and benchmarks obfuscate the real performance of models. Checkout our new work (comprehensive survey…
There’s no singular benchmark for this, but LLMs are shockingly bad at negative instructions. Don’t say the word delve ➡️ DELVE Avoid using emojis ➡️TONS Never use punctuation ➡️ TONS Here’s to hoping future LLMs understand what I don’t want as much as what I do want.
🚨 Effective Altruism's Bait-and-Switch: From Global Poverty to AI Doomerism 🚨 The Effective Altruism founders planned – from day one – to mislead donors and new members in order to build the movement's brand and community. aipanic.news/p/effective-al…
Boeing is literally the textbook case study for regulatory capture. Which is exactly where we're heading with the new AI proposed regs -- big guys stifle competition and we are left with just regulatory theater.
@natfriedman Automotive legislation sidebar allowing our teams to evaluate legislation in realtime rather than having plough through numerous regionally specific documents. Massive value add, super simple given the expanded context window,..
@natfriedman find patterns in and across your data continuously
@natfriedman make explainer animated videos from a description/script
@natfriedman a tool that can read and understand a large complex existing codebase and suggest ways to improve security, performance, maintainability (by humans and AI agents)
@natfriedman Not sure what you would call it exactly but maybe something like “User directed censorship” for entertainment…. So you pick Home Alone to watch with your 8 yr old but want to bleep out the bad words. There is so much entertainment with bad language that I’d love to simply…
@natfriedman Asynchronous codebase reviewer. Have some agent recursively combing through files in my repo, finding places to add abstraction or clean up. Present me with findings each morning. I hate writing code and realizing I wrote a helper for that 2 years ago lol
@natfriedman research agent that runs in a loop refining its answer for 2-3 hours. Would pay like $100-200 per search. Am probably going to make this
@natfriedman "Ghost employees". Take a top performer that has left, use all of their slack/email/intranet history and create a useful remote employee chatbot for all the shit that didn't get KT'ed in time. Ethics of it get weird if the employee is still alive.
@natfriedman semantic search of the inbox. for searches like this:
@natfriedman I’m looking for a product that proactively searches my recent browsing and other history and helps re-surface relevant resources as I work Could be part of a broader platform I describe in this 1-pager Can’t find it! docs.google.com/document/d/1in…