I legitimately dont know how to reply, bc by this point llms co-own all aspects of my life and jumps between gpt4->claude3->claude3.5->o1 have all been very noticeable
I'm the opposite. We're presumably in a similar line of work, but while I've experimented with every major release from OpenAI and Anthropic last year -- I've barely ever used an LLM outside of that.
I still Google things I want to know and skip the AI part.
> I still Google things I want to know and skip the AI part.
My Google use is down significantly. And I mostly reach for it when I am looking for current information that LLMs do not yet have training data for. However, this is becoming less of an issue as of late. DeepSeek for example has a lot of current data.
the problem is that last generation of the largest models failed to overcome smaller models on the benchmarks, see lack of new claude opus or gpt-5. The problem is probably in the benchmarks, but anyway.
regarding the prompt processing and token generation you are correct
it makes sense to benchmark them infependently since prompt processing is done in parralel for each token and is compute bound and token generation is sequential and bound by memory banwidth
If someone wanted to make the case that Facebook is a net-good force in the world, I'd read that. But I haven't seen anyone even try to lay out evidence of Facebook doing any good things. I like pytorch, prophet, and their decent job policing the pedophiles on the FB platform, but what else do they do that's good and responsible?
Years ago, when FB was relatively new, I was very anti-FB, and would discourage everyone from using it due to privacy concerns.
Then a Sudanese friend was staying with me for a few days. He had spent years separated from his relatives who lived in multiple countries, as he was working odd jobs to save enough money for a degree. Through FB, he got to see his nephews and nieces be born and grow for a few years. The amount of joy it brought him was immeasurable.
I realized I was the asshole thinking I knew better than he. If it brought him this much joy, it definitely was worth the loss of privacy. Multiply that by millions.
Today, there may be viable alternatives, but there weren't in those days. At least none that his non-techie relatives could use. FB is what brought about the joy. Nothing else did, despite trying.
FB is a tool. Yes, there are plenty of problems with it, but they require mitigations - perhaps even legislation - not a shutdown. Killing FB, IG and Whatsapp really won't solve anything. Plenty of competitors will eat up the space. It's like saying "Marlboro is the market leader. Let's ban it."
Their product kills people or makes them seriously ill - how would you improve that product so it doesn't?
So for the good of society - access to their products should be much harder, i.e. factor in the cost of all the damage those products cause and people will have less of an incentive to purchase these.
There was, and the reality is that most people were not able to use it effectively enough to be a replacement for FB. Sure, tech heavy users were fine with mailing lists, cc's, attachments, etc - but most of the world wasn't.
Email's been around forever. If email were a viable alternative, FB would not exist. The amount of sharing between the Sudanese guy and his family skyrocketed after FB came along.
Email connects 4 billion people and helps small businesses reach customers. And email manages to do this without starving local journalism, or algorithmically recommending extremist groups to susceptible people, or cause genocides, or release SOTA creepshot Ray-Bans, or feed COVID disinformation to my parents every day.
I think facebook tries to keep people locked into their platform so the users believe this exact lie. Facebook doesn't enable the capabilities of the internet.
Why is anything good objectively?
It is good for me because I can stay connected with my mom despite living in different country.
Also facebook group for my town is nice and useful, and my wife enjoys posting in instagram from our trips.
Can you support that belief with any evidence or a coherent argument? What good does FB do that offsets the great amount of evil it does?
What offsets FB's choices that connected millions of extremists and gave them protected spaces to radicalize each other? [0]
What offsets FB's decision to rush into regions it can't moderate, leading to FB being used as the platform that spreads genocide-inciting misinformation? [1]
Personally I've had to work very hard to keep my parents safe this pandemic, disabusing my parents of dangerous COVID related misinformation that they saw on Facebook. I've had to research some really deranged stuff that they picked up on that platform.
I see a pretty deep debt on the evil side of the ledger, but you're asserting FB is net positive. So what am I missing? Where is the good stuff FB does that offsets all of this harm?
This warrants a long response, that I would love writing one day but don’t have the time.
But I’ll leave a short one here to respect your comment.
I believe social media empowers people, it makes my life better, it gives many people a voice where before it only belonged to a few. Those few also incited wars and genocides, and bred rage and mistrust (From NY Times supporting Iraq war to your local news stations to Fox/CNN concentrating on stories that outrage or warm your heart and get you back) from before Facebook existed and do so to this day. All of those nasty effects of social media existed before, they just moved to the most efficient media form.
I support people’s right to communicate and own their opinions, and I accept that giving a voice to everyone will always results in problems small and large. But Facebook does spend a lot of resources trying to make social media better. I’m an insider and swayed, but I think Facebook spends much more resources than TV channels and newspapers trying to address such issues.
The rest is a proper response really depends on a lot of arguments clumped against Facebook, and addressing them depends on the person:
- Is the criticism against all social media?
- Is it against ranking feed items?
- Is it against monetizing through ads?
- More rarely, is it against censoring, or because of data policies, or supposed negligence, or more.
There’s such so many issues against Facebook, yet almost everyone keeps using the products, and the only countries that ban Facebook are not ones that come of as inspiring to me when you consider their reasons.
"I think Facebook spends much more resources than TV channels and newspapers trying to address such issues."
Proportional to their extremely greater profit they better be.
"All of those nasty effects of social media existed before, they just moved to the most efficient media form."
Back in the day you could kill a few people with arrows and spears, then you had automatic rifles, now we have nuclear and chemical weapons. We actually try to regulate those...
I stopped using Facebook in 2017. I felt absolutely miserable while angrily reading the stream of ragebait Facebook kept feeding me. Facebook has been really damaging to the mental health of my parents which has been an absolute nightmare for me over the past two years. I went through my dad's feed last year too see if FB had improved and it was an absolute hellscape. There would be nice content like grandbaby pictures from his friends sprinkled between lethally dangerous COVID disinformation and content designed to enrage. Much like what was captured in this article [0].
I don't see the value of giving a platform to dangerous bullshit. People need truth about reality. Giving a platform, or voice, to people who feed bad information to others can cause those others to make suboptimal choices, like rejecting masks or vaccines. How many people have died a miserable death (suffocating because their lungs can no longer absorb any oxygen even at 100% O2) over the past 2 years because of COVID disinformation spread on FB? Based on how many hours I've had to spend on the phone with my parents explaining why things they saw on FB are wrong, there's absolutely no way it's less than 10s of thousands just in the US.
I'm pretty impressed with your ability to just brush aside a genocide caused by FB rushing into a new market with 18 million people with only a few dozen moderators that speak the language. Facebook had been running a social media site for nearly a decade by then; there's no way FB didn't know the minimum ratio of moderators to users needed to keep up with typical moderation workloads. The only explanation consistent with the evidence is that FB just didn't care about avoiding that genocide and mass displacement. And based on your reaction to it, I assume the FB culture still isn't concerned about the boring details of being responsible and ethical.
I'm not trying to shame you, I don't think shame works at changing people's behavior, but I do think being caviler about massive harms like genocide and COVID disinformation is a major red flag, and I think you should take a hard look at your values and grapple with the possiblity that working for a company that kills hundreds of thousands through (best case scenario) gross negligence may be indefensible.
Nothing is objectively good, good and bad are just words we invented to describe our preferences. My preference is that people don't continue down the path of intense tribalism, but other people obviously can have different preferences.
Text, email, zoom, etc. Facebook did not invent online communication. Facebook didn't invent the ability to share photos online.
As far as I can tell, Facebook's innovation is it's great at identifying people susceptible to extremist conspiracies and at connecting them. Per Facebook's own research:
"""Even before the teams’ 2017 creation, Facebook researchers had found signs of trouble. A 2016 presentation that names as author a Facebook researcher and sociologist, Monica Lee, found extremist content thriving in more than one-third of large German political groups on the platform. Swamped with racist, conspiracy-minded and pro-Russian content, the groups were disproportionately influenced by a subset of hyperactive users, the presentation notes. Most of them were private or secret.
The high number of extremist groups was concerning, the presentation says. Worse was Facebook’s realization that its algorithms were responsible for their growth.The 2016 presentation states that “64% of all extremist group joins are due to our recommendation tools” and that most of the activity came from the platform’s “Groups You Should Join” and “Discover” algorithms: “Our recommendation systems grow the problem.”"" [0].
There were QAnon groups with millions of members. Facebook created these echo chambers and then ushered people in. And the only thing you can think of is they allowed you to communicate with your mother and share images?
Logic from converse for one. If it isn't good would it be good to cut off communication despite such a power being heavily abusable? If not then why is this point a magical perfect balance of communication?
Lol, you're talking as if it's unique to Facebook. Anyone with money and a modicum of technical expertise can make a messaging platform. The hard part is monetization.
Most people in the world actually don't use Facebook for messaging. Messenger is second to WhatsApp.
There are dozens of software that can do worldwide instant messaging. It's ridiculous to think that that's why Facebook is successful. It's not, monetization and network effects are the reason, not being technically superior.
Hi, I worked there for over a year and am familiar with the specific research discussed in the article.
IG is a net good because even in this specific case, it did more good than harm. The numbers in the article are all less than half. On average users self-report increased well being when they use IG, as long as they don't use it too much.
But improving things on average isn't good enough. IG also wants to improve the situation for the users who self-report being adversely affected by their IG usage. IG has added features and spent a ton on research and product improvements to improve subjective wellbeing, and have been successful in measurably improving self-reported wellbeing, reducing objective measures like bullying prevalence and negativity.
