Papers of the day   All papers

The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence

Comments

Gary Marcus: "absolutely brilliant" —Nobel Laureate Danny Kahneman The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence It's what I wish I had had time to say at the #AIDebate :) Finally ready, free, on arXiv. Happy Reading! https://arxiv.org/abs/2002.06177

31 replies, 1144 likes


محمد الحامد: ورقة بحثية نشرت قبل يومين، استمتعت بقراءتها و انصح المهتمين بالاطلاع عليها #الذكاء_الاصطناعي #تعلم_الآلة "The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence" https://arxiv.org/pdf/2002.06177.pdf

1 replies, 229 likes


Florian Huber: Two very knowledgeable articles about current shortcomings in #AI/#ML and promising ways towards more general, robust, and flexible #AI: 1."On the measure of intelligence" @fchollet: https://arxiv.org/abs/1911.01547 2."The Next Decade in AI" @GaryMarcus: https://arxiv.org/abs/2002.06177

1 replies, 119 likes


Pasquale Minervini: "Recent work by Minervini et al (Minervini et al., 2019) gives me hope that a neuro-symbolic hybrid approach could break new ground. [..] The very fact that people are trying gives me more hope; reasoning and knowledge need to be first-class citizens if we are to move forward" ❤️

1 replies, 111 likes


MONTREAL.AI: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence "absolutely brilliant" —Nobel Laureate Danny Kahneman @GaryMarcus : https://arxiv.org/abs/2002.06177 #AIDebate #ArtificialIntelligence #CognitiveModels #Reasoning #RobustAI https://t.co/kAe0RxnwW6

1 replies, 94 likes


MONTREAL.AI: "absolutely brilliant" —Nobel Laureate Danny Kahneman The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence Gary Marcus: https://arxiv.org/abs/2002.06177 #AIDebate #DeepLearning #RobustAI https://t.co/R64DN5qTes

0 replies, 92 likes


Guy Van den Broeck: Happy to see @GaryMarcus mention statistical relational learning, probabilistic programming, and probabilistic circuits as promising directions for hybrid AI approaches: https://arxiv.org/pdf/2002.06177.pdf (also see https://www.youtube.com/watch?v=mQMxqecxGhk)

0 replies, 75 likes


Felix Hill: I just don't get why classical electronic computers are such a pervasive source of inspiration for cognitive scientists (see article cited below) It feels to me like a biologist proposing that the girders, joists and windows in skyscrapers must be core components of physiology

14 replies, 72 likes


Erik Brynjolfsson: A cool new paper shows how Neural Networks can speed up simulations by billions of times. https://www.sciencemag.org/news/2020/02/models-galaxies-atoms-simple-ai-shortcuts-speed-simulations-billions-times Is this because deep learning leverages the simplicity of the laws of physics, per @tegmark et al. ? https://arxiv.org/pdf/2002.06177.pdf #AI #2MA #MachineLearning https://t.co/hsjEimjJKS

4 replies, 63 likes


JohnMark Taylor: The fields of Cog Sci and AI were born joined at the hip but gradually drifted apart; if these clearheaded and interesting proposals are on the right track, then the reintegration of these fields might be one of the most exciting things happening today.

3 replies, 62 likes


Carlos E. Perez: Gary Marcus latest manifesto "The Next Decade in AI" https://arxiv.org/abs/2002.06177 attempts to reintroduce the philosophical foundations of GOFAI (i.e. knowledge, internal models, reasoning) into the current AI conversation.

3 replies, 62 likes


Gary Marcus: Yes! Reading these two papers together, comparing and contrasting and seeking your own synthesis, would be a great way to think about the road to a more robust AI.

1 replies, 56 likes


Luis D. Duval: A great article by @GaryMarcus about the future of Artificial Intelligence. Brilliant conclusions, and implications. https://arxiv.org/pdf/2002.06177.pdf #AI #TechnologyVision #ArtificialIntelligence https://t.co/yw1MiIdPpc

2 replies, 50 likes


hardmaru: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence @GaryMarcus https://arxiv.org/abs/2002.06177 https://redd.it/f5kh7v

1 replies, 43 likes


Dagmar Monett: "[A] hybrid, knowledge-driven, reasoning-based approach, centered around cognitive models, that could provide the substrate for a richer, more robust AI than is currently possible." 👏 That's @GaryMarcus' The Next Decade in #AI: Four Steps Towards Robust Artificial Intelligence

1 replies, 41 likes


Charan Ranganath: This paper brings up many points that are relevant to understanding episodic memory. Cog Neuro has a fetish with event boundaries (I.e. context shifts) but we really need to understand event structure. (10 bucks says it doesn’t live in the hippocampus or entorhinal cortex)

7 replies, 37 likes


Gary Marcus: puzzled about what symbol-manipulation is and how it might relate to reasoning? today might be the perfect day to read “The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence” https://arxiv.org/abs/2002.06177

4 replies, 34 likes


Dr Catherine Breslin: The next decade in AI - should focus on hybrid solutions, incorporating structured knowledge, abstract reasoning, and rich cognitive models of the world (from @GaryMarcus) https://arxiv.org/pdf/2002.06177.pdf

4 replies, 32 likes


Dorothea Baur: Great input from @GaryMarcus: Robust #AI needs "deep understanding": the ability not only to correlate & discern subtle patterns in complex #data sets, but also the capacity to look at any scenario & address q's such as who, what, where, why, when and how. https://arxiv.org/abs/2002.06177

4 replies, 32 likes


Geoffrey Siwo: Need to rethink AI, big data alone is not enough for true intelligence. We need knowledge-driven, reasoning based approach @GaryMarcus https://arxiv.org/pdf/2002.06177.pdf

3 replies, 31 likes


Luis Lamb: State-of-The-Art: GNNs meet Neural-Symbolic Computing https://arxiv.org/abs/2003.00330 w/@vardi @AvilaGarcez @Melleo54Sis @marceloprates_ in the light of @GaryMarcus 's The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence https://arxiv.org/abs/2002.06177 https://t.co/v1vQ1QTLM2

0 replies, 30 likes


Mehmet Suzen: Any research decision-maker should read @GaryMarcus 's works. Anyone tries to silence him, is doing a tremendous disservice to the AI community. We need down to earth cognitive scientist contributing to the development of new AI technologies. Thank you Gary.

1 replies, 30 likes


reza mahmoudi: Great read on the next decade in #AI: "Four Steps Towards Robust Artificial Intelligence" by @GaryMarcus https://arxiv.org/pdf/2002.06177.pdf #DeepLearning #machinelearning #ArtificialIntelligence #AI https://t.co/ygC4OAF4Ay

0 replies, 28 likes


Fabian Westerheide: Reading now "The Next Decade in #AI: Four Steps Towards Robust Artificial Intelligence" by @GaryMarcus #RiseofAI read it yourself => https://arxiv.org/abs/2002.06177 https://t.co/ffQtYimJJo

1 replies, 23 likes


Niloofar: An amazing must-read! "When sheer computational power is applied to open-ended domain—such as conversational language understanding and reasoning about the world—things never turn out quite as planned." My favorite quote!

0 replies, 22 likes


Elisenda Bou: "Recent research in AI and ML has largely emphasized general-purpose learning and ever-larger training sets (..) I propose a hybrid, knowledge-driven, reasoning-based approach (..) that could provide the substrate for a richer, more robust AI" @GaryMarcus https://arxiv.org/pdf/2002.06177.pdf

2 replies, 22 likes


Bennie Mols: Great read on the next decade in #AI: "Four Steps Towards Robust Artificial Intelligence" by @GaryMarcus Free on arXiv: https://arxiv.org/pdf/2002.06177.pdf

0 replies, 19 likes


Luis Lamb: Great points raised by @GaryMarcus : happy to see the promising perspectives of neural-symbolic computing in AI,which has been an active field of research @AvilaGarcez @luislamb @pascalhitzler @jeublanc @marceloprates_ @Melleo54Sis @vardi @frossi_t @erikbryn @kerstingAIML

1 replies, 16 likes


Alex Hamilton: "A growing body of evidence shows that state-of-the-art models learn to exploit spurious statistical patterns in datasets… instead of learning meaning in the flexible and generalizable way that humans do."

3 replies, 16 likes


Eyad Nawar: A great article by @GaryMarcus. A must read.

1 replies, 16 likes


Dwayne R. Monroe is mixing a martini in your honor: I've been spending time reviewing, studying and visualizing @GaryMarcus' paper in arXiv titled "The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence" - https://arxiv.org/abs/2002.06177 /1

1 replies, 15 likes


Srinath Srinivasa: From deep learning to deep understanding. Robust AI as an achievable next goal for AI, rather than the lofty AGI. Wonderfully written article!

1 replies, 15 likes


miriam meckel: Worth a read ...

1 replies, 15 likes


Luis Lamb: #AIDebate @ceobillionaire

0 replies, 13 likes


Yariv Sadan: Great essay on gaps with deep learning techniques and the progress we need to develop more robust AI.

0 replies, 13 likes


Luis Lamb: “The next decade in AI: Four steps towards robust Artificial Intelligence” by @GaryMarcus is on arXiv: https://arxiv.org/abs/2002.06177 #robustAI #AIDebate #hybridmodels #neuralsymboliccomputing @AvilaGarcez @RealAAAI @vardi @ceobillionaire @Montreal_AI

0 replies, 13 likes


Luis Lamb: This is really worth reading, if one is interested in AI and its future, by @GaryMarcus

0 replies, 12 likes


Andrew Revkin: This is only the beginning of efforts to use #AI to break longstanding climate-modeling barriers. Here's a glimpse of the ferment (hat tip @PierreGentine): https://twitter.com/apilipis/status/1219356507864616962

0 replies, 12 likes


Kristian Kersting: Thanks for your vision @GaryMarcus, even before it actually went online. It was a great pleasure and honor to provide my humble feedback. Very interesting and important directions for #AI and #ML

0 replies, 12 likes


Ioana Ciucă: Hello, weekend read. 👌

0 replies, 12 likes


Jarno M. Koponen: Towards hybrid models. @GaryMarcus defines the fundamental building blocks in his latest paper: “we should seek to build learning systems that start with initial frameworks for domains like time, space, and causality, in order to speed up learning” https://arxiv.org/abs/2002.06177 #ml #ai https://t.co/nZqWhTcMnB

1 replies, 11 likes


Jarno M. Koponen: Thinking: ”rather than starting each new AI system from scratch... we should seek to build learning systems that start with initial frameworks for domains like time, space, and causality, in order to speed up learning” -@GaryMarcus https://arxiv.org/abs/2002.06177 #ml #ai https://t.co/1l89uIVvcx

0 replies, 10 likes


HotComputerScience: Most popular computer science paper of the day: "The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence" https://hotcomputerscience.com/paper/the-next-decade-in-ai-four-steps-towards-robust-artificial-intelligence https://twitter.com/GaryMarcus/status/1229216106377646080

0 replies, 10 likes


David Broadhurst: This article by @garymarcus is well worth a read for anyone interested in the future practicalities of effective #AI. But, I can't help but feel that the most productive (rapid) way to improve the general perception is to stop using the term #AI for all #ML systems - misleading.

0 replies, 9 likes


Gary Marcus: https://arxiv.org/abs/2002.06177

0 replies, 9 likes


Stefano Marzani: Must read! #AI #ArtificialIntelligence

0 replies, 7 likes


Vilynx: 'The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence' by @GaryMarcus is a must-read! > http://arxiv.org/abs/2002.06177 #AI #ML #DL #CognitiveComputing #MachineLearning #neuralnetworks https://t.co/15lhc951Sz

3 replies, 7 likes


Dr Jean-Louis Amat: Alternative ways than using deep learning for any #AI problem are needed. This paper from @GaryMarcus should be read if you are interested in the future of artificial intelligence.

1 replies, 7 likes


BBVA Data: Interesting paper by Gary Marcus: 4 Steps Towards Robust Artificial Intelligence

2 replies, 7 likes


MIT Technology Review: #EmTechDigital https://arxiv.org/abs/2002.06177

0 replies, 6 likes


Michael Harries: Highly recommended - a clear eyed view of the challenges facing deep learning and why incorporating symbolic methods matters. AKA - what's next for AI.

2 replies, 5 likes


Carter Maslan: Challenges for the next decade of AI summarized in one image https://arxiv.org/pdf/2002.06177.pdf via @GaryMarcus https://t.co/9NTkmMTedT

0 replies, 5 likes


Laha Ale: An excellent paper!

0 replies, 4 likes


Gary Marcus: @zaidalyafeai @NaxAlpha @xsteenbrugge see also my discussion of Meena in my Next Decade article. https://arxiv.org/abs/2002.06177

2 replies, 4 likes


kordoni_nlp: Great read on the next decade in #AI: "Four Steps Towards Robust Artificial Intelligence" by @GaryMarcus Free on arXiv: http://arxiv.org/pdf/2002.06177…

0 replies, 3 likes


Ariel Meilij: Reading this Monday

0 replies, 3 likes


Leo Brueggeman: Follow up from @GaryMarcus on the #AI debate (http://shorturl.at/twzU9) TLDR is that #DeepLearning should incorporate symbols that can be manipulated (eg serval present in photo is a _TYPEOF_ cat) to become more robust https://arxiv.org/abs/2002.06177 #machinelarning

0 replies, 3 likes


Gary Marcus: @missy_cummings @AnimaAnandkumar my suggestions here: https://arxiv.org/abs/2002.06177

0 replies, 3 likes


Marga Robles: 👉The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence https://arxiv.org/abs/2002.06177

0 replies, 3 likes


Phil Harvey: Best paper I have read in years ... awesome. And likely to be proved very important #AI

0 replies, 3 likes


Beril Sirmacek 🦋: Such a good read from @GaryMarcus Thanks @ProfLavesson for the suggestion! https://arxiv.org/abs/2002.06177

0 replies, 3 likes


Jen LaPlante: I read this over my holiday. A great follow-up to "Rebooting AI" by @GaryMarcus. If you have an interest in AI, I highly recommend this read. #ai https://arxiv.org/abs/2002.06177

1 replies, 3 likes


José A. Alonso: The next decade in AI: Four steps towards robust Artificial Intelligence. ~ Gary Marcus. https://arxiv.org/abs/2002.06177 #AI

0 replies, 3 likes


AI Papers: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence. http://arxiv.org/abs/2002.06177

0 replies, 2 likes


Nick 🇪🇺: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence https://arxiv.org/abs/2002.06177

0 replies, 2 likes


Valentino Zocca 🇮🇹 🇪🇺: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence - https://arxiv.org/pdf/2002.06177.pdf #DeepLearning #MachineLearning #AI By @GaryMarcus

1 replies, 2 likes


Pascal Marco Caversaccio: Anyone truly interested in #AI, that means no Chief Bullshit Director, should read @GaryMarcus's paper on "The Next Decade in AI". https://arxiv.org/abs/2002.06177 For the guys who think they are too busy, you can find a great summary here: https://www.linkedin.com/pulse/next-decade-ai-gary-marcus-50-pages-my-summary-fabio-ciucci/?trackingId=os%2B5YFDcQWC8%2FfpcJANTjw%3D%3D

0 replies, 2 likes


Val: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence by G. Marcus https://arxiv.org/abs/2002.06177 #AI #MachineLearning #ArtificialIntelligence https://t.co/my6FAIgpk4

0 replies, 2 likes


Abel.TM: Great to read a broad analysis about the future of AI by @GaryMarcus, including evaluations of current technologies, their limitations towards robust AI and alternative directions of research https://twitter.com/GaryMarcus/status/1229216106377646080

1 replies, 2 likes


═╬╬═: The Next Decade in #AI: Four Steps Towards Robust Artificial Intelligence #deeplearning https://arxiv.org/pdf/2002.06177.pdf https://t.co/paDgkz70Db

0 replies, 2 likes


thinkofx: Excellent summary of the limitations of Deep Learning

0 replies, 2 likes


Agnese Chiatti: "Anybody who closely follows the AI literature will realize that robustness has eluded the field since the very beginning." #AIdebate #whatnext #robustAI #intelligence

0 replies, 2 likes


Benoit Hudzia: If it is written in Python, it's probably machine learning If it is written in PowerPoint, it's probably AI If it works in production, it's probably an If/else statement

0 replies, 1 likes


Vilynx: A good read for this weekend: 'The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence' by @GaryMarcus > http://arxiv.org/abs/2002.06177 #AI #ML #DL #CognitiveComputing #MachineLearning #neuralnetworks https://t.co/lTAQoNYsXb

0 replies, 1 likes


John Clark: [2002.06177] The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence I’ll give it a try. https://arxiv.org/abs/2002.06177

0 replies, 1 likes


Daniel Muller: Me

0 replies, 1 likes


Tim Clancy: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence https://arxiv.org/abs/2002.06177

0 replies, 1 likes


Greg Weinger: Fascinating thoughts on the future of AI by @garymarcus The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence Downloadable PDF - https://arxiv.org/abs/2002.06177 http://bit.ly/38N3b6n

0 replies, 1 likes


Arnab Sanyal: Interesting !!

0 replies, 1 likes


Dr. Rajeev Bukralia: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence https://arxiv.org/abs/2002.06177 #AI #ML #DeepLearning #DataScience #Research

0 replies, 1 likes


Yunus Şahin: It was an excellent read and full of insights which can also be useful for my future MSc and PhD study. Thanks @GaryMarcus!

0 replies, 1 likes


Dr. Rajeev Bukralia: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence https://arxiv.org/abs/2002.06177 #AI #ML #DeepLearning #DataScience #Research

0 replies, 1 likes


John Grant: "Within a few years, I predict, many people will wonder why deep learning for so long tried to do so largely without the otherwise spectacularly valuable tools of symbol manipulation; ..." — Gary Marcus https://arxiv.org/abs/2002.06177 h/t @blopeur

0 replies, 1 likes


Jibu Elias: A must read for anyone into #AI, by ⁦@GaryMarcus. ⁩ The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence https://arxiv.org/abs/2002.06177

1 replies, 0 likes


Kai: Great paper by @GaryMarcus! I experimented with NNs to solve automatic music genre classification (paper coming soon). Came to a similar conclusion as Gary. SOTA systems lack basic cultural knowledge. Without this, they are is easily fooled. 1/2

1 replies, 0 likes


Max Todaro: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence https://arxiv.org/abs/2002.06177

1 replies, 0 likes


Anita Yorker: [2002.06177] The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence https://arxiv.org/abs/2002.06177, see more http://tweetedtimes.com/v/1407?s=tnp

1 replies, 0 likes


Pascal Marco Caversaccio: @GaryMarcus cites in his recent paper (https://arxiv.org/abs/2002.06177) a funny #Meena conversation snippet that actually confirms deep learning's stupidity (the original source can be found here: https://github.com/google-research/google-research/blob/master/meena/meena.txt):

1 replies, 0 likes


Jordi de la Torre: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence https://arxiv.org/abs/2002.06177

1 replies, 0 likes


Steve H ☕️: 13/ (There's a connection to AI in there: @GaryMarcus juxtaposes abstract/symbolic understanding with deep learning/intuitive understandings, and argues that true intelligence requires both) https://arxiv.org/abs/2002.06177

1 replies, 0 likes


Content

Found on Feb 17 2020 at https://arxiv.org/pdf/2002.06177.pdf

PDF content of a computer science paper: The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence