Nvidia is sued by authors over AI use of copyrighted works
abhibeckert @ abhibeckert @lemmy.world Posts 0Comments 1,096Joined 2 yr. ago
abhibeckert @ abhibeckert @lemmy.world
Posts
0
Comments
1,096
Joined
2 yr. ago
How would it be moderated and by whom?
That would be easier to answer if we had a list of companies that can afford to buy it (that's a short list) and also willing to buy it (an even shorter list).
I don’t necessarily think forcing them to sell to another entity will fix the problems
Sure - it obviously depends who buys it. Elon Musk, for example, would probably be a bad steward.
But what about Alphabet? That might not be so bad. As a fan of YouTube, I'd love to see the "shorts" feature killed off and all that content moved to a separate service where I can go the rest of my life without ever seeing a short repeating video.
Whoever buys it, it the US can force TikTok to be sold once, they can do it again if the buyer proves to also be problematic.
They're not worried about CSAM. They worried about TikTok users being influenced during an election campaign.
And yes, it is a moderation issue. Specifically, the US doesn't want the current moderation team to be in charge of moderation.
Disclosure: I don’t use Facebook, Intagram, Twitter, nor TikTok
To put it in perspective, about a quarter of the US population uses TikTok. And politics are a major discussion point with the political content you're exposed to selected by an algorithm that is opaque and constantly changing.
It absolutely can be used to change the result of an election. And China has meddled in elections in the past (not least of all their own elections... but also foreign ones:
"China has been interfering with every single presidential election in Taiwan since 1996, either through military exercises, economic coercion, or cognitive warfare, including disinformation or the spread of conspiracies"
logging statuses that were typed out, but ultimately not posted.
That's common practice across the web. For example in a lot of social networks (not Lemmy) if I were to close this reply box without clicking the "Reply" button, I'd simply be able to open it and my half written reply would be restored.
It's not about gathering data on users, it's about saving users from accidental deletion particularly on touch screens where it's so easy to accidentally brush the wrong button.
Also I'd argue anything you intentionally type into a facebook status message box, is something you're happy to share with facebook and everyone on your social network. There's no expectation of privacy and we self sensor what we type into that box.
Logging key strokes is different, there's no reason to record that anything you record (other than actually typing messages) is likely to be something the user intended to type somewhere else but didn't realise which browser window had keyboard focus.
They would be picking up all kinds of things including passwords.
For example — last year they sent what a spy balloon over the USA in what a lot of experts believe was a test of US defence systems - weapons used to shoot down the balloon had never been used before outside of top secret test facilities. And that balloon was covered in high tech sensors and almost certainly broadcasting data in real time. There's no plausible explanation for the incident other than to find out how the US would respond.
Why does China want to know how US defence systems work? A lot of people already think there's a chance of war between the two super powers. That balloon incident didn't help things.
And what went viral on TikTok? Claims that the balloon was a actually flying over Canada and never went near US soil. Claims that it was launched by kids in the USA. Where did those claims originate from? Nobody knows, but it seems pretty coincidental. These claims were spread on other social networks too - but they went viral on TikTok alone.
That's not the only incident, it's just one of the most recent one that involved TikTok. Others have been far more serious especially in busy international waters south of China.
If Twitter's financial backing by Saudi Arabia/Qatar is ever a concern, I'm sure the US will act on that as well.
Your description doesn't seem to match what the site does? For example the front page has a function that converts uppercase text to lowercase text.
It's not article content - it's an interactive utility.
Not sure what it's like in Darwin, but here in QLD all of the old 3 bedroom homes have been renovated with their verandah/patio/etc converted into bedrooms. Often increasing them from 3 bedrooms to 5 bedrooms.
And if they're on stilts... people raise the homes up and add an entire new home - might now have 9 bedrooms in a home that used to be 3. You can rent that downstairs living space out for 500 bucks a week so it easily pays for itself.
If the house can't be renovated, you can sell the unobtanium old growth rainforest timber for a pretty decent price then build a large modern concrete home in it's place.
The average american home is 2,000 square feet. The average Australian home is 240 square metres (2,600 square feet).
Keep in mind America has cities with half the population of Australia and people living in those cities are not living in huge homes.
customers say X, they probably mean they want Y and Z
Sure - an LLM can help catch some of those situations. But if anything it makes prompt engineering even more important.
Sometimes the customer actually wants X, and a prompt engineer needs to predict this issue and disable the Y/Z behaviour. Prompt engineering is changing, but it's not going away.
Future systems could for example start asking questions more often
Current systems already do that. But they're expensive and it might be cheaper to have a human do it. Prompt engineering is very much a thing if you're working with high performance low memory consumption language models.
We're a long way from having smartphones with a couple terabytes of RAM and a few thousand GPU cores... but our phones can run basic models and they do. Some phones use a basic LLM for keyboard auto correct for example.
I really wish all of these companies racing to replace their existing software features and employees with LLMs understood this.
They totally understand it. And OpenAI has solved it. For example while researching The Ultimate Answer to Life the Universe and Everything, I asked it to calculate 6 by 9 in base 13 and got the correct answer - 42.
ChatGPT didn't use the LLM to calculate that. It only used the LLM understand an obscure and deliberately confusing chapter of the Hitchhiker's Guide book, to write and execute this python script.
# To calculate six by nine in base 13, we multiply the numbers in our standard decimal system and then convert the result to base 13.
# Calculate 6 * 9 in decimal
result_decimal = 6 * 9
# Convert the result to base 13
# The easiest approach is to use the divmod() function repeatedly to get the remainder (which corresponds to the base 13 digit)
# and update the quotient for the next iteration until the quotient is 0.
def decimal_to_base_n(num, base):
if num == 0:
return "0"
digits = []
while num:
num, remainder = divmod(num, base)
digits.append(str(remainder))
return ''.join(digits[::-1])
# Convert the decimal result to base 13
result_base_13 = decimal_to_base_n(result_decimal, 13)
result_base_13
If it's like my seat, the gas seal is probably stuffed. I should probably fix it but I just treat it as a reminder to stand up.
API requests are usually encrypted with SSL and protected against unauthorised use with something along the lines of a JWT: https://jwt.io/
Breaking through the SSL might be possible, if the developer doesn't pin certificates, but you don't know the secret used to generate the HMAC signature (blue section of that website), then you can't simulate the API request. And the secret shouldn't be sent over a network connection.
You could probably access the secret with enough work, but it would be a lot of work. You'd have to do it separately for each app. And the developer can change the secret whenever they want. The developer will change the secret at the slightest hint of anything like this being used with their app. And possibly also take additional steps to keep it from being accessed (e.g. store it in the Trusted Platform Module or equivalent on Android/iPhone). Even the CIA can't access that - it's mostly intended for payment processing and protecting data on a stolen phone, but there's nothing stopping a weather app from using it to prevent unauthorised access to their API (weather data is very expensive, and often billed per API request).
Running the real app on a real phone though... basically nothing an app developer can do to stop that.
Train infrastructure is being removed around the world - good luck convincing people to build more.
The fact is a train turns one trip into three trips - truck to the railway station, train to another station, truck to the final destination. That often adds days to what otherwise might be a 3 hour delivery - because trains are only cheap if you send about a hundred or so trucks full of cargo on a single trip.
Only really makes sense for really long trips but more and more of those are done by ship or airplane. Trucks aren't going anywhere.
It's the battery, and the support frame to carry the weight of the battery safely. Like it or not - cars are getting heavier.
The code is obviously protected by copyright. Not sure why anyone would question that?
If the prompt is protected, then the output image will be too (and by the same owner). I suspect it depends on how detailed the prompt is (just like a tweet might not be eligible for copyright, unless it's a particularly creative joke/haiku/etc).
As a seasoned camper the original design doesn't look even remotely weatherproof. So to me, this is an improvement.
The lack of tent mode (suspension self levelling) is disappointing but in reality other vehicles don't have that feature either and the world is full of level campsites. It's also not that hard to manually level things up with ramps or a jack.
More disappointing is there's no way to connect the tent to the HVAC system - that would have been a massive selling point if they could manage it and surely wouldn't have been difficult? Just need some sort of opening or pipe to the main cabin and a few software tweaks to make sure you don't drain the battery overnight.
I think you’re wrong about government funded news. It will always be less biased than privately funded news because democratic governments (even Trump) are held accountable in a way that no private company ever will be.
We definitely shouldn’t get all our news from the government but it’s hardly the horror show Americans seem to think it is. Especially in a democratic country.
Facebook was not built on the backs of journalist has always been about chatting to friends and family first, strangers second, and news a distant third
Facebook has clearly stated they don’t profit off news. I’m inclined to believe them until proven otherwise, especially since I can open the Facebook app and there is literally no news anywhere to be seen.
All of the news where I live is paywalled.
Why should Facebook pay for content that users cannot even access? Surely anyone who is actually paying for access isn't finding articles from Facebook - they are getting the news directly from the source.
The only content that isn't paywalled is from ABC, but their definition of "local" news is national news which happens to be a story about something that happened here. Usually reported on so poorly it's obvious the journalist lives a few thousand kilometres away and doesn't really understand the issue at all.
The state of journalism in Australia is really bad - but I don't think Facebook paying Murdoch did anything at all to fix things. If anything news is even worse now than it was few years ago when the deal was originally signed.
The government needs to go back to the drawing board and re-think their approach. Personally i'd like to see something similar to our health system where we have a national journalism budget dedicated to funding private journalists the same way Medicare funds businesses in the medical industry.
And make sure there are strict rules around ethical journalism and also encourage original reporting - zero funding if you re-hash news broken by someone else.
Apple has always allowed side loading just not as a method of mainstream distribution.
App developers use it all the time and not just on their own devices, it's also used to beta test apps.
There area few different ways it can be done with varying limitations. But in general it is absolutely possible to install apps outside of the App Store. In fact side-loading has been around since before the App Store even existed.
Almost anyone (except Tim Sweeney) can register for a developer account. And in fact you don't even need to, since most people who side-load are testers not developers. Developers can deploy a pre-release build of their app to up to ten thousand regular iPhone owners bypassing the app store entirely.
I have no idea what they're talking about with "only 3 apps". As an iOS developer I've got countless side loaded apps on my iPhone. Some of them are test builds of my own apps, some of them are test builds of other people's apps, some of them are hobby projects where I've slapped together an app in a weekend — not good enough for public distribution but it works for me and I've been using them for years.
Give it time. The DMA came into effect today. And it includes provisions for the European Commission to measure how effective it is and supplement the DMA with orders that only apply to individual companies without applying to the entire industry.
It’s more like charging the iron ore mining companies over gun murders.
NVIDIA doesn’t have any say over how their GPUs are used.