do you think it possible to solve misinformation with AI?

Question

Arsen Misakyan · Accepted Answer

@nicole_ogloza Sure thing! Misinformation is a big PITA nowadays.

Arsen Misakyan · Answer

I use GPT-3 quote a lot. Mainly content related stuff. Judging from what kind of good job it currently does, I would definitely vote YAY.

It took a long time for ANI(Narrow AI) to become AGI(General AI). AI itself, collectively speaking, is learning and growing exponentially. IMHO we're just couple of steps away from ASI(Super AI; level: god).

TL;DR: AGI just needs the proper training to sort out "good" and "bad" info.

Samir Moussa · Answer

I used to work in the AI media space. Fake news and entity recognition was the biggest issue we had at the time. AI needs training data and that only comes from human fact checkers. Because of the nature of news, you need to manually fact check new information and feed it as training data. So half of the work is done using humans with AI being used for aggregation and pattern recognition. I do believe we will have AI solving all problems, including this, but not right now.

Fares · Answer

I have doubts that we can solve misinformation with AI. Because if we talk about misinformation, it means that the information has already been assimilated by a person and it is very difficult to fight this cognitive bias with impersonal action.
On the other hand, if it is to locate the sources of disinformation it is simpler, but the drift is dangerous.

John · Answer

I've created AI modules for 99% of my work.

One project that I created was an Algo trading system, divided into 3 modules.

Module A, is a newsreader that processes what it reads, then submits the ticker symbol or company name to module B.

Module B, processes info submitted by module A. Review real time ticker symbol trades and see the stock's reaction to the news in the next few minutes. When it sees a clear direction, up or down, makes a decision to buy or sell a stock. Submits order to Module C.

Module C, places the order to the open market and submit order confirmation back to module B for real-time account balance, etc.

On a very basic level, for me to identify fake news, I would depend heavily on well known stock market finance news websites to minimize fake news.

Then my AI will track news reports and will determine if the source contradicts the market action and take it from there.

I've been creating my own AI systems for more than 30 years to assist me with my daily IT job functions.

I usually train my AI first, and watch it make its own decisions to test it, before implementing the AI module.

Qudsia Ali · Answer

There are enough AI algorithms available to challenge fake news, and more are being developed in time. Unfortunately, that doesn't mean that we can sit back and relax. Although these technologies can help curb the miss information campaigns, they could also launch a new era of fake news and online misinformation.

Nicole Ogloza · Answer

@arsen_misakyan thank you for this! Super helpful. We are currently trying to do this for the news media and track patterns of how misinformation 'escapes' happy to share more :D

Dylan Merideth · Answer

also the embodiment of the solution is going to be pivotal to adoption, not sure what it would look like, but it will have to have nuance

Dylan Merideth · Answer

I think there is an implementation of a sufficiently advanced AI fact checker that comes and helps this. But the tricky thing is freedom of expression and garnering societal consensus for a tool that aims to address misinfo.

do you think it possible to solve misinformation with AI?

Replies