Top all-time · #1059
Piekniewski's blog
Personal blog · blog.piekniewski.info · on HN as felippeeaiphilosophy
Piekniewski's blog is a personal blog at blog.piekniewski.info about ai and philosophy. Since 2016, 66 different Hacker News readers have linked it 116 times, most often “AI winter is well on its way”. It is still writing: the last post was on 2024-11-13, and you can follow it by feed.
Most-linked posts?
- AI winter is well on its way9 people · 10 mentions · first linked 2018-05-30
- AI winter – update6 people · 6 mentions · first linked 2018-10-29
- Farcical Self-Delusion5 people · 5 mentions · first linked 2021-12-19
- AI Mid 2021: Self Driving Car Meets Reality5 people · 5 mentions · first linked 2021-05-12
- DeflAition: Artificial Intelligence Deflation of 20204 people · 4 mentions · first linked 2020-04-14
- Deep Learning – The “Why” Question (2018)4 people · 4 mentions · first linked 2018-10-13
- AI the No Bullshit Approach4 people · 4 mentions · first linked 2020-06-08
- AI Circus Mid 2019 Update3 people · 3 mentions · first linked 2019-05-30
- What would an autonomous car do?3 people · 3 mentions · first linked 2017-01-24
- AI Update, Late 20193 people · 3 mentions · first linked 2019-11-18
What Hacker News says?
In a sense, I would say yes there are learning laws, but it's still early in codifying them. Along one axis, you could compare: supervised, semi-supervised, self-supervised and unsupervised learning. Along another axis, consider that there are versions of each method that take into account temporal/dynamic data, versus others that require randomly shuffled static data. In the current problems of visual perception, I…
Comment linking /2016/11/01/statistics-and-dynamics/
on “OpenAI Universe”Submitted AI winter is well on its way as a story · 993 points
HN discussionWhat you mention in point #1 isn't the main issue with games, but with the "reactive" approach in AI. Reactive AI systems are necessarily going to be more limited than say, predictive AI systems (a new emerging class of AI models). [1] Agreed, though, that the focus on games is a long-standing source of problems holding back AI. My colleague has referred to this as "ludic AI" in a recent blog post where he expands o…
Comment linking /2016/11/30/learning-physics-is-the-way-to-go/
on “OpenAI Universe”Submitted Autopsy of a deep learning paper as a story · 249 points
HN discussionThere is a never ending confusion caused by the term "AI" to begin with. Term coined by John McCarthy to raise money in the 60's is really good at driving imagination, yet at the same time causes hype and over-expecations. This field is notorious for its hype-bust cycles and I don't see any reason why this time would be different. There are obviously applications and advancements no doubt about it, but the question…
Comment linking /2016/11/17/myths-and-facts-about-ai/
on “This AI Boom Will Also Bust”Submitted AI Update, Late 2019 as a story · 138 points
HN discussion> People used to say this about every single thing that computers can do better than people. But computers can't do it better than people! That's what drives me nuts about this debate -- it's just accepted as a premise that either the self-driving cars are much safer than human drivers, or the path to getting them there is very close and no serious obstacles remain. Neither is true and it's not clear they will be. h…
Comment linking /2017/05/11/a-car-safety-myths-and-facts/
on “Uber shuts down self-driving trucks unit”Submitted Elon and the collective as a story · 93 points
HN discussionAI Circus, Mid 2019 Update https://blog.piekniewski.info/2019/05/30/ai-circus-mid-2019-... Stories of AI Failure and How to Avoid Similar AI Fails in 2019 https://www.lexalytics.com/lexablog/stories-ai-failure-avoid... How IBM Watson Overpromised and Underdelivered on AI Health Care https://spectrum.ieee.org/biomedical/diagnostics/how-ibm-wat... Automakers Are Rethinking the Timetable for Fully Autonomous Cars https…
Comment linking /2019/05/30/ai-circus-mid-2019-update/
on “Arend: Theorem Prover Based on Homotopy Type Theory by JetBrains”Submitted AI Mid 2021: Self Driving Car Meets Reality as a story · 85 points
HN discussion- nl · 2020-06-24
Here's a good example from 2018: "Deep learning (does not) scale... We can't just scale up AlexNet and get respectively better results - we have to fiddle with specific architectures, and effectively additional compute does not buy much without order of magnitude more data samples, which are in practice only available in simulated game environments." https://blog.piekniewski.info/2018/05/28/ai-winter-is-well-o... Bu…
Comment linking /2018/05/28/ai-winter-is-well-on-its-way/
on “GPT-3” Submitted Deep Learning – The “Why” Question (2018) as a story · 81 points
HN discussionThat reminds me of deep learning. Investment in computing resources is definitely increasing exponentially over time: https://openai.com/blog/ai-and-compute/ https://techcrunch.com/2019/11/10/the-post-exponential-era-o... Tthere doesn't seem to by any evidence of exponential gains in performance, even on their own benchmarks (which importantly don't measure "intelligence", but something more narrow): https://blog.pi…
Comment linking /2019/11/18/late2019-the-wizards-of-oz/
on “Mark Zuckerberg Interviews Patrick Collison and Tyler Cowen”Submitted How close are we to solving vision? as a story · 65 points
HN discussionSo Lex Fridman curated this page and is the subject of many videos within it. Seems like Missy Cummings and Filip Piekniewski's assumption that MIT has scrubbed him from the site is unfounded. [1] https://twitter.com/missy_cummings/status/117949700363566285... [2] https://blog.piekniewski.info/2019/11/18/late2019-the-wizard... Lex's blocking of dissenting opinions on twitter is still egregious though.
Comment linking /2019/11/18/late2019-the-wizards-of-oz/
on “MIT lecture series on deep learning in January 2020”Submitted Singularity Missed as a story · 35 points
HN discussionYes, the field has emerged out of MNIST/ImageNet and that is what those algorithms are optimised for. For modelling actual dynamics different design is necessary. It happens that the design that makes sense also seems to agree very well with the observed biology of the cortex. You can find links to our Predictive Vision Model in this thread as well as a few additional thoughts here: http://blog.piekniewski.info/2016…
Comment linking /2016/11/30/learning-physics-is-the-way-to-go/
on “Predictive Learning [pdf]”Submitted AI Update, Late 2020 – Dumpster Fire as a story · 25 points
HN discussionHi, I'm the blogger. I just want to add a very simple statement: In order to create a model of the world, the machine learning substrate has to have the capacity to cover the observed dynamics. Hard not to agree with this, almost sounds like a tautology. Now let's try to derive conclusions: - World dynamics is full of multi-scale interactions (e.g. in vision illumination of a single pixel depends on the whole scene…
Comment linking /2016/11/04/predictive-vision-in-a-nutshell/
on “Predictive Learning [pdf]”Submitted AI winter – update as a story · 21 points
HN discussionLecun has identified a real problem for AI -- the need to understand the real world, the link between intelligence and prediction over time. But the tools he is using are not the right ones. Deep conv nets were not designed with prediction over time in mind. Here's one reason: Deep convolutional nets do not handle dynamical information from their lowest layers. By design, conv layers and pooling layers immediately b…
Comment linking /2016/11/21/yann-lecun-cmu-11-2016-comments/
on “Predictive Learning [pdf]”Submitted Recurrent neural networks, dreams, and filling in as a story · 14 points
HN discussionYou should probably reconsider your trust in the face of data: https://blog.piekniewski.info/2018/02/09/a-v-safety-2018-upd... But who cares about data, if it is all about AI singularity religion?
Comment linking /2018/02/09/a-v-safety-2018-update/
on “The Battle for Best Semi-Autonomous System: Tesla Autopilot vs. GM SuperCruise”Submitted AI winter is well on its way (2018) as a story · 12 points
HN discussionIt depends on the context. Humans may make a better decision if there is something else providing the information. Perhaps the manhole cover sits at the curb triggering immediate attention. Perhaps there are cones, only in the wrong place because some drunk guy moved them for fun. Such is reality. The problem is not so much availability of information (a sensor packed car has a lot more information than a human behi…
Comment linking /2016/11/03/reactive-vs-predictive-ai/
on “Toyota's Gill Pratt on Self-Driving Cars and the Reality of Full Autonomy”Submitted DeflAition: Artificial Intelligence Deflation of 2020 as a story · 8 points
HN discussionI took the liberty to draw a few situations which AI would have likely no clue what to do, even if it detected the danger. I'll be adding some more as every day on the freeway brings me new ideas (in fact I'm editing one right now). Something like a "Winograd Schema for self driving cars": http://blog.piekniewski.info/2017/01/19/what-would-an-autono...
Comment linking /2017/01/19/what-would-an-autonomous-car-do/
on “Toyota's Gill Pratt on Self-Driving Cars and the Reality of Full Autonomy”Submitted A good critique of Musk and some of his more math-challenged ideas as a story · 7 points
HN discussionI commend you. Very few people actually try feeding images into these things to show how bad they are. For a real shock at how poor performance is, try feeding frames of video in. (Video is different because it generally doesn't have carefully stereotypically framed, exposed, in-focus content). Edit: examples of ResNet applied to video by a colleague http://blog.piekniewski.info/2016/08/12/how-close-are-we-to-...
Comment linking /2016/08/12/how-close-are-we-to-vision/
on “Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017””Submitted A Brief Story of Silicon Valley's Affair with AI as a story · 6 points
HN discussion
