Stay up to date on all things crypto and blockchain
Token Daily is a place to discover trending news and products in crypto and blockchain.
Source: techcrunch.com ·
Forecast, a new project from Facebook’s internal R&D group, NPE Team, is launching today to build a community around predictions. The iOS app will allow users to ask questions and then use in-app points to make predictions about what might happen in the future. Users can also create, discuss and view these crowdsourced predictions. At […]
Jeff Garzik - @jgarzik
2 weeks ago
RT @FEhrsam: Facebook wades into prediction markets. Smart.
While crypto will be the substrate for prediction markets over time, good chance they start as a game with social status and evolve into real money later.
h/t @dwr
https://t.co/iYGOOaH5ta
12
85
Balaji S. Srinivasan - @balajis
2 weeks ago
RT @Mehdiyac: Interesting new product by @Facebook
Forecast is an app to build community around crowdsourced predictions.
Will it help find the truth?
It would be great to push people to have skin in the game and bet on their predictions the way @balajis suggested.
https://t.co/tHt217Ubba
26
161
Fred Ehrsam - @FEhrsam
2 weeks ago
Facebook wades into prediction markets. Smart.
While crypto will be the substrate for prediction markets over time, good chance they start as a game with social status and evolve into real money later.
h/t @dwr
https://t.co/iYGOOaH5ta
8
39
waterloo302
2 weeks ago
Perhaps, uness there was aggregation of prediction tx's (e.g. by a proxy) into batches to keep within main chain scale capacity.
1
MadDoctor5813
2 weeks ago
At least Rehoboam made its own predictions. This is just asking people what they think will happen.
2
blammycuzzing
2 weeks ago
I’m predicting many “settled” answers from people denying science’s facts regarding COVID.
Also- “At a time when Facebook and other major tech platforms are under fire for their role in aiding the spread of misinformation, fake news, propaganda, conspiracy theories and other non-factual content presented as truth, an app focused on making “guesses” about the future seems ill-advised...”
2
cr125rider
2 weeks ago
We are publishing opinions of people who don’t know stuff? Cause more Facebook?
3
spongyruler
2 weeks ago
Facebook creeped me out this morning. Last night before bed, I had only thought about needing lotion because my skin is super dry. I didn't say anything out loud, didn't do any searches anywhere, and this morning as I'm scrolling through, I see an ad for lotion. I think I saw 2 ads total for lotion, so it easily could've been a coincidence, but it was still weird. Facebook needs to calm down.
2
drcopus
2 weeks ago
https://en.m.wikipedia.org/wiki/Confirmation_bias
There is absolutely no way that Facebook targeted those ads at you for those reasons. Definitely a coincidence, but you may fall into a demographic that is more likely to be given ads for lotion (the connections the system makes can be complex), and hence the coincidence is less spooky than it may seem.
3
theKoala_man
2 weeks ago
Listening to Triple J (Australian Radio Station) the other morning and they were talking about the new World's Fastest Computer in Japan. Basically Rehoboam...
Nolan & Joy are incredible... check out Person of Interest if you haven't already. And then you'll look at the world we live in now very differently.
2
MouldyTaco2
2 weeks ago
People get to vote on what they think is going to happen and whichever outcome has the most points is awarded "most likely." Far shot from Rehoboam.
5
drcopus
2 weeks ago
Has anyone here actually read this article? It's basically just a crowd sourcing platform.
This isn't AI
17
drcopus
2 weeks ago
You know that supervised learning isn't the only form of learning, right? Labelled datasets are just a really bad way of defining a utility function.
To provide a bit of credibility to my opinions, I'm a PhD student in AI. A Rehoboam-level AI is not going to be constructed from a labelled dataset of human predictions about such complicated matters. The data would be much too sparse for the problem domain. The system would need to be totally constructed before the data was useful (i.e. complex priors are needed to make inferences from data like this).
2
Marksman79
2 weeks ago
Yeah, I have a high level understanding of the most common types of ML being used today, but certainly not to the fidelity a researcher like yourself must have. Although I was making a tongue in cheek joke implying that a simple labeled dataset would be enough to make an entire predictive AI (as if all you need to control the world was a more advanced captcha-based network), since I have you here: what's your general thoughts about the self driving long tail problem we're facing today? Will the approach Tesla has talked about (automatically labeling video, closed loop fast iterations) be fruitful, or could another system approach the asymptote more quickly?
1
drcopus
2 weeks ago
> I was making a tongue in cheek joke
Fair enough, sorry if I responded aggressively
> what's your general thoughts about the self driving long tail problem
The specifics of self-driving cars are not my speciality at all, but I have worked with autonomous control and computer vision. Also, as someone who doesn't drive my intuitions about the cognitive capabilities needed for the task may be underdeveloped.
Anyways, in my view there are three major challenges involved with autonomous vehicles:
1. **Perception**: Taking data from your sensors and extracting information, e.g. geometric understanding of your surrounding, entities such as pedestrians, and instructions from road signs.
2. **Planning and Simulation**: Taking your perceptions and simulating what will happen as a result of various actions, e.g. being able to anticipate where a pedestrian will go next and what your brakes do.
3. **Adhering to Norms and Laws**: Understanding the "rules of the road" is probably a lot more difficult that is immediately obvious. There are the easy rules, such as don't drive on pavements or hit pedestrians, but there are also unspoken norms and "common sense" that is particularly hard to program into a computer (or have a computer learn).
Machine learning has been particularly helpful for perception, and traditional AI techniques have given us sufficient tools for planning and simulation (at least within the context of autonomous vehicles), however, I think the third challenge is where we are still struggling.
This also helps explain why AVs are so effective on motorways, where the rules are well-understood and consistent, but aren't so effective in urban areas. I think that it may seem like we are very close to self-driving because the majority of driving is like motorway driving, so when you see stats it says that 90%+ of the trip is self-driven. However, we were able to do that in specific conditions in the 90s, so progress on the last 10% has been remarkable slow.
All that being said, my intuition is that the level of commonsense reasoning needed to close the gap for driving isn't that much, and we probably will be there in 5-10 years.
2
Eriugam31
2 weeks ago
Wouldn't surprise me if Zuckerberg was in fact an AI of some sort, he doesn't show any emotion or regard for other human life.
3
vismundcygnus34
2 weeks ago
I’d bet it already exists somewhere on a smaller scale, and was probably utilized in the 2016 election.
5
dangerouspitcher
one true thing 2 weeks ago
And that is how you came to be, mon ami. The Zucc's minions brought you into this world, and have ensured that you will outlive us all, that the strategy will succeed.
3
ChthonicPuck
2 weeks ago
That's not bad. Since we were on the topic of Westworld I thought of Dolores. It got it wrong, guessed someone from The Vampire Diaries, but she was the second runner up prediction!
Edit: Take 2, got it right. Guessed Lisbeth Salader from The Girl With The Dragon Tattoo.
1
danystark1
Do they dream? 2 weeks ago
O my god, yesss. I haven't played that in years. But true. This IS the REAL all knowing AI
6
smashingdon
2 weeks ago
I already deleted Facebook off my phone and have separate browser for FB on my pc. I don't use google for searching nor chrome, but I'm using Android. But I dont use my phone that much. Probably time to get off gmail as well I guess
3
forgotmapasswrd86
2 weeks ago
Unless you just use landlines and send paper mail for communication, theres no way to stay off some company's database.
4
676f626c7565
2 weeks ago
So all of us sane people who ditched fb and their products a while ago are going to end up in freezer racks in the desert arent we?
199
26_Charlie
2 weeks ago
Unless you live in a country with "the right to be forgotten" laws, Facebook never stopped collecting data on you.
3
blackstars321
2 weeks ago
Doesn't really matter if you ditched Facebook or not. They steal your information from all the people you know who still use it. Plus there's always still Google. Who is just as evil these days and has even more information on us than Facebook.
16
bmwbryan
2 weeks ago
You and me both, pal. I just binged all 3 seasons in the past week or so and I didn’t realize it got cancelled... so imagine my surprise when I got to the end of season 3.
2
SeSSioN117
2 weeks ago
Are we the same person? That's literally what happened to me. I was actually so happy that Westworld picked up a few elements from Colony. The moment they started speaking about "Outliers" I started smiling.
inb4 the warning Solomon was gonna give Caleb was not about mankind's extinction being caused by themselves or the hosts but something else. Ayy either way, I'm happy with where Westworld is going.
I kept thinking to myself, it's entirely possible Colony is a parallel universe to Westworld, or maybe it was just another "world".
1
Trending this week..
Newsletter
Upcoming Q&As
Post a Job
Teams Hiring Now
-
TokenSoft is the volume leader in compliant token sales.
-
The open protocol for tokenized debt.
-
A secure online platform for buying, selling, and storing digital currencies.
-
A second layer, off-chain scaling proposal for bitcoin.
-
Ensuring the blockchain is inexpensive and accessible to everyone.
-
An open protocol for decentralized exchange on the ethereum blockchain.
- See all