On Thursday, the fourth, some institutions from Bulgaria and Quatar along with MIT have been working on an AI that is able to distinguish whether or not an article is credible and biased or not. Using 1,066 sources, the AI distinguished 65% of factual sites correctly and 70% of biased sites right. It even used the articles sources to make a decision. A person with a postdoctoral degree working on the project, Ramy Baly, stated that "It turns out that Wikipedia is important." and that having a Wikipedia page is telling of its trustworthiness. However, Wikipedia is not the most reliable of sources, but it is also approximately stated in the article that it can say things about a news source that may not be obvious to an algorithm like this, like the Onion being satiracle. The AI was also able to look at 50-150 articles on a news site and check the language used in writing it to see if it was biased. Baly said that biased sites normally have a more emotional appeal to the audience, and have a different writing style when compared to down the middle sources.
How accurate do you think something like this AI could be in the future? Would you ever consider using it?
Personally, I think it might work to some extent, but I would double-check my articles first. If a more capable or sentient AI is created, then I believe that it would be more possible to get the accuracy on a fact-checking AI to a higher percent than 65%. At this point, it might just be easier to double-check sources than rely on an unreliable machine. When it gets more testing to the point of higher accuracy, then I might use it.
If this AI is ever heavily relied upon, do you think that producers could alter the way data and articles are interpretted to the AI so unreliable sources or biased sources are passed off as correct? Do you think that anyone would realize it?
Having the AI altered seems like a possibility, but not a big deal. I think sites would still be up and running, and the curious people would still double check things. It might mess with the people who don't dig a tiny bit deeper, though. It would be kind of like fake news, maybe looking real with nothing else supporting it.
Replies
Great job Amelia!
I think it has a possibility of working, but only to extent. However, I would definetly want it to be a more advanced AI before I trust it fully, even then I'd still be a little skeptical. I feel like AI can be very helpful, but only if the producers can alter the way the data and articles are intepretted.
I think this might work, but there will probably be some glitches, and it's not always going to be correct as it only distinguished 65% of factual sights right and 70% of biased sights right. Also, Al could be very helpful in the future, but only if the producers can alter the way data and articles are interpretted. I feel like that's the only way it's going to work right.
I believe that this Al could be benefital and I could consider myself using it a ton because I do not know how many times I try to research something and it ends up being false information about the topic. This Al sounds like a great idea and could help a lot of people get the right correct information to learn, grow, and would be faster in the long run!
I agree, if this works it seems like a good place to find sources. I can relate to not being able to find good articles on my topic, and this seems like a huge help. Unbiased news would probably allow more people to find their own idea on things, instead of being lead by strongly left or right sided articles. It would also reccomend a better description of what actually happened, instead of showing off tilted numbers and stories.
In the future, I think it could definitely be more accurate, but since it's only in its early stages right now we can't know for sure. I would definitely consider using it for projects that require credible sources, or to use in debates where facts are needed. I think that the producers could definitely do that, but considering how quickly people on the interenet spot deception they would be found out eventually.
I wouldn't quite rely on it yet, it isn't exactly the most accurate of things at the moment. Like you said, we'll just have to wait and see how this thing works out.
The internet does have a huge amount of conspirists and sleuths, but then again, there was a petition against banning water that actually gained signatures. It called water by its atomic name, and had some loosely-correlated evidence. People aren't always the smartest with news, and some people believe everything they hear. There definitely are people who would be able to find out the producers' scheme, though.
I personally do not think that something like this would be very accurate because it might not be able to tell if there is a link taking them to a seperate site showing something completely opposite of wha they are saying on the page the AI checker is checking.
I do believe that the producers could alter the data and articles whether or not they will get caught and get fined heavily is just the luck of the draw.
It might be able to pick up on external links eventually, though there probably would need to be some extra work. It would probably be hard to distinguish outside sources from other articles from the same publisher, though.
I don't think nobody would notice because cetain people decide to double check everything. They would probably be a tiny minority, though. The producers would probably get fined because humans could double check the AI's work later.
If it were perfected I'm sure it would be n amazing peice of technology that would greatly help our time but not so until its ready to do so. If the technology was working and someone found a way around it im sure people woouldnt notice because it was reliable when they first used it.
-
1
-
2
-
3
-
4
of 4 Next