ericjang

Humor me - what would you use such an arm for? What does the design look like? How many times a day would you use it? A lot of time in homemaking (a full time job for some people) is spent cleaning, unpacking things, and packing things away. How would such a system work?

This happens all the time at Google, though the mundane details of internal tech churn are not often publicized. Source: am Google employee.

I think that's probably the way to go if a business wants to tackle the problem of a useful home robot in the next decade. I totally agree with your point about home developers being forward-looking and designing homes with robots in mind.

That said, I don't relish the idea of my oven or toaster being internet-connected (https://tech.slashdot.org/story/20/11/25/1910244/aws-outage-...).

I also think it's good to also think long term beyond what is immediately feasible or commercially viable. Personally, as someone who enjoys the sci-fi movie Blade Runner, I would much prefer to see a robotic android being "more human than human", than having to compromise in the design of homes.

are you sure that would happen? Make no mistake, I get that Western powers do have cultural and economic agendas (McDonald republics and such). But if you look at Japan and South Korea, they didn't really have significant protests or any "installed leaders".

This poem is dismissive of the progress humanity has made in computing and information technology. Tools like Pandas and Arduino and computer graphics were not mature in 1970s, and in order for data science, hobbyist electronics, or modern filmmaking to get where it is today, it was built on a massive ecosystem of CRUD and poorly designed abstractions like OOP.

But it is still an enjoyable read, and contains kernels of truth - we have abandoned the unix philosophy and created bloated, poorly engineered software. I think the comparison to music and hair cutting is very apt - the mark of a skilled engineer is knowing how to limit scope and ambition so that value can be delivered quickly and elegantly.

We all roll our eyes at the next baby sitter app, but at the end of the day we probably use some app to find a baby sitter - whether it be Google Search or Yelp. In a biological ecosystem, all things that can be, will be.

Thank you, dang! Having to read so many pages of ceaseless bickering and techno pedantry must take a toll on the psyche. I appreciate all that you do!

It is widely believed by neuroscientists that spike timing in animal brains (whether by aggregate rate or by individual spikes) does the majority of information encoding.

Curious, why does this specific arxiv submission keep on cropping up popularly over the years? Is this common for any top post?

The report, while a fun thought exercise for any aspiring academic, is not a novel insight, and is obviously untrue for most definitions of market efficiency.

If we are going to play the academic one-upmanship game, a more general result that "best" or "multiple" N-player Nash Equilibria for N > 2 is already NP-hard. The implication of an efficient market would be if every player had a polynomial-time algorithm to solve the NP-hard problem, ergo P=NP.

[1] https://people.csail.mit.edu/costis/simplified.pdf

[2] https://www.quantamagazine.org/in-game-theory-no-clear-path-...

[3] https://arxiv.org/abs/1104.3760#:~:text=Unlike%20general%20N....

thanks, I was able to figure it out.

If you highlight the name, the highlighted text is {{14}}. I don't think the lead investor's name is necessarily 6 characters. FWIW, I was deeply curious about who the investor was and spent some time investigating who lead the fundraising rounds on Crunchbase, but I couldn't narrow it down sufficiently.

I searched for the "famous paper" in question and could not find it - can you point us to a link? So far I find two of Dijkstra's talks/writing on anthropomorphism

[1] https://www.cs.utexas.edu/users/EWD/transcriptions/EWD05xx/E...

[2] https://www.cs.utexas.edu/users/EWD/transcriptions/EWD09xx/E...

Sorry, I misread the original question. I am not aware of the CCP accessing TikTok user data (for instance, to track dissidents or surreptitiously use it to compute social credit scores).

Yes https://twitter.com/ericjang11/status/1289713401992683523?s=...

What are some examples of this?

Same technological challenges, but the rendering of past historical figures' ethnographic features has much higher social stakes and is a considerably more politically sensitive issue than how a T-Rex is rendered.

Thanks Daniel for your response! I hope my GP comment didn't come across as hostile or discouraging to future artists who want to explore this kind of art. I want to be clear that I'm not offended by your work.

My discomfort stems more from my belief that a non-technical / layperson audience is much more likely to conflate "each step towards realism" with a "step towards ground-truth", the exact opposite to what you suggested. The writeups make clear when information is not available, but the artistic medium itself (a rendered still image showing only one possibility) does not showcase this.

> In response to Cocci’s findings, Voshart removed all mentions of the site and revised several portraits to better reflect their subjects’ probable complexions

You graciously made corrections for plausibly Nazi historical propaganda. However, not every artist/historian may elect to do this, out of scholarly pride or a personal politics or just laziness. The artist subsequently wields tremendous power over the images that show up on Google Image search, and their decisions may unintentionally contaminate future historical research by way of inducing subtle prejudices.

We live in an age of misinformation - and I think the information that must be protected most carefully is information of the past, specifically around people. DeepFakes of living celebrities are fairly easy to verify and disprove, but as soon as you go back ~5 decades or so, some people start to challenge whether certain events (Holocaust, Tiananmen Square Massacre) even actually happened.

Some ideas on how one might mitigate this:

- Provenance technology that shows the editing history (potentially including a MD5 of the dataset) of how an image was synthesized.

- A simple, unobtrusive watermark on the center of the image indicating that it is but one of N possibilities, where N gives a rough estimate of the entropy over the distribution. Sort of like the recycling triangle numbers.

- GIF form showing multiple possibilities given the data you have.

> On a completely different note, I am not sure what you’re getting at with the racial stuff

What I am saying is that this sort of work can lead to people believing these images as true, i.e. "this is what Roman Emperors looked like". If you ask an ML system to solve a problem for which the inputs cannot possibly determine the answer (i.e. the color of a fossilized animal or the skin color from a person's statue), it will give you a highly biased answer.

I fear that some laypeople think of ML as "magic software that can do anything" without considering whether a problem is non-ambiguous given its inputs (e.g. something like object classification is less ambiguous than this problem).

I'm not saying "burn it to the ground" - I wanted to learn more about what data and model was used to generate these images so that I might understand better where potential sources of racial / ethnographic bias might come from.

> trying to find racial undertones in every tangential thing

If you read my comments history you'll see that I'm no critical race theorist, nor do I discuss everything in a racial context.

Out of curiosity, what "cause" are you talking about?

This is undoubtedly cool, but I worry quite a lot about people who aren't educated about ML interpreting these "historical renditions" as a "ground truth forensic". Rendering fossilized animals and coloring black and white photographs is also cool, but when you start to render the bone structure and "caucasian-ness" of statues, it starts to veer into some uncomfortable racial and ethnographic territory.

What data and models were used to synthesize these predictions?

Out of curiosity, what would such a model be used for? I worry that as ML is becoming more popular and powerful, people are jumping to use it on problems from which the answer cannot possibly be identified accurately from the inputs alone. This model's predictions would entirely based on historical trends of what images looked like, rather than something objective like carbon dating.

I am not discouraging someone from building such a model, but it would be really helpful to know the context for which such a model is being developed. If it's just a hobby investigation, it would be cool to see how "predictable" dates are from images. I could even see it being used in forensics to provide a "first guess" as to when an image occurred, and helping with triaging of evidence. However, things become deeply problematic if the result from the image is fed to people as "ground truth" simply because the model was found to be accurate on a validation dataset. I certainly wouldn't want this model to be used to determine whether a suspect is innocent / guilty, or to be used naively by museums to date photographs.

> You can connect them just like you do with micro services or modular systems - what is the issue about this?

Clearly I'm missing something here. Can you walk me through how exactly a micro service would enable GPT-3 to look up facts, and incorporate that knowledge into a conversation? What would the microservice API look like? How are the outputs consumed?

I feel that overestimation of fundamental capability is actually the cause of many morality/safety issues, if not at least a degradation of user experience (automated voice menus).

The big difference here is that TikTok, Twitter, Whisper actually have working technology, in spite of ethical concerns. What I am saying is that the people who want to use GPT-3 for business use case X, Y, and Z probably have not thought deeply enough about the limitations/implications of large language model methodology on specific nuances of X, Y, Z tasks.

Have you considered the fact that GPT-3 can't actually look up any information? Consider the implications of that before suggesting that GPT-3 could be used for therapy.

Fair enough, thanks for your reply and clarification. I think "PR nightmare" scenario might be more likely than you suspect.

1. What if an AI therapy bot tells a depressed person to kill themselves? How do you get a language model to obey confidentiality rules? How do you prevent it from memorizing and regurgitating someone's mental health conversation to another patient?

2. I think the implications of replacing journalists with far cheaper automated systems (with substantially less fact-checking capability) have not been well thought out, and I worry that some VC bro is going to rush a product to market before policy makers / stakeholders have thought carefully about whether this is something that we want.

It's telling that GPT-3's best writing successes have been of the "philosophical musing" variety, not of writing accurate articles. I'm not sure whether that says more about AI or Philosophy.

The author's main points are all centered around the assumption that GPT-3 related technologies are ready to create market-disrupting products, and that the key business challenges remaining are mostly around figuring out how to build a defensible moat.

However, that assumption is dead wrong.

We don't know how well GPT-3 "works out of the box" (hence OpenAI's api release), and it's stupendously premature to assume that it is ready for products.

We don't have a good understanding of what inputs the model handles well vs. doesn't (a key pre-requisite for building a safe product), and it has been shown to regurgitate authoritative but untrue statements.

[EDIT]: by construction, it also cannot store information provided by the user for later use, nor can it "look up" facts.

I agree with the high level statement ("starting a business around gpt-3 is a bad idea") but almost none of the generic business advice, because it's founded on an assumption of technical capability that simply isn't there yet. The article talks about GPT-3 in an abstract "super powerful AI" sense, which leads me to suspect that the author doesn't really understand the technical limitations of GPT-3 beyond the demos that have been shown.

It's less of a matter of "I don't want the government to spy on me, the individual with a banal life" than a matter of "I don't want the government to spy on us, a society".

For instance, having a FB account enables FB to spy on non-FB users via shadow account - they will create shadow accounts based on faces they don't recognize in photos, and shadow friend profiles based on numbers stored in my phone.

A more insidious situation is that TikTok enables Bytedance (read: the CCP) to monitor political dissent and potentially shape cultural zeitgeist via content recommendation [1]

Many content recommendation algorithms use some sort of collaborative filtering algorithm, where if the system lacks data on one user's preferences they can infer it given a sufficient number of users with overlapping preferences and attributes [2]. So even though my data and life are not special, it does enable the government to find out more about people they are interested in.

[1] https://www.theguardian.com/technology/2019/sep/25/revealed-... [2] https://en.wikipedia.org/wiki/Collaborative_filtering

"The Man Who Solved the Market: How Jim Simons Launched the Quant Revolution" discusses the lineage of Shaw, Tartaglia, and his predecessors briefly. I think some more academic texts on Pairs Trading and Stat Arb will mention the PDT team as well.

I read somewhere long ago that there have been incidents of FBI employees tasked with reviewing child pornography [1] who saw so much that they became desensitized or even aroused by it. Does anyone know if this is well-documented, or was this an exception rather than a common occurence?

[1] https://www.fbi.gov/history/famous-cases/operation-innocent-...

In the process of learning about a family of algorithms in machine learning I also gained some physical intuition of determinants (same diagram as 3Blue1Brown, but applied in a different context of "squashing and stretching" probability mass): https://blog.evjang.com/2018/01/nf1.html

This is horrific. When did this happen? Was this in Harlem?

I think the basic premise is that the status quo of transactions (free service for user data) has resulted in an imbalance of power for tech companies, and that by increasing the cost of acquisition, users have a means to collectively negotiate better "market rates" from the tech companies for services rendered. It can be regarded as similar to "unionizing" the users.

It would be interesting to get some preliminary guidance on what tech companies would do in response (start charging for premium services? only give premium services to the most valuable users?)

Andrew Yang has also mentioned in prior podcasts on UBI (Freakanomics, etc.) that probably the best way to fund such a program would be proceeds from increased taxation on tech companies.

I for one would love to see cable companies turned into public utilities. But that goal is not mutually exclusive from the goal to give the consumer some more power over the big tech conglomerates.

Disclaimer: I work at Google, opinions my own.