Human vs AI content, can we tell the difference anymore? AI technology like ChatGPT, DALL·E 2 and QuillBot AI are blurring the lines between what content is created by a computer or humans. It's also complicating the way we consume content as you now need to be suspicious about who or what really created what you are consuming.
Over the past several years the rise and rapid innovation in the AI space is making it difficult to predict what the future of content creation will look like. There’s plenty of stories about it in the news. The sotries about how AI is impacting education, business and social media is intriguing.
source
For example there was a recent article that really got me thing about how Students are using ChatGPT to cheat and how complex and difficult it has been for teachers and school districts to combat AI. From my perspective, AI has its place and should be embraced by teachers, business and content creators to more effectively complete task. However, if we blindly use AI as a substitute to actual research, content creation we are doomed.
I recently ran an experiment on Hive with AI. I used ChatGPT to create three blog post on three topics to see if anyone could actually tell the difference between my content and AI generated content. Those 3 posts were the following:
- Hive Blockchain Overview
- Super Bowl Ads Will Look Different This Year
- Jeffersons Very Small Batch Review
I was a little bit apprehensive about performing this experiment for the blowback it might receive. However, I thought the conversation around this was worth the risk. CHatGPT made writing these post scary easy. It took about 5 mins a post to write the prompt and let ChatGPT to create and organize the content. After doing a quick proof read to try to catch and glaring errors I left the rest of content to try to preserve the content how it was presented to me. After that I added a few of my photos and hit the post button.
These post received about the same engagement as my other post with my core group of commenters stopping by. These posts also produced about the same amounts of rewards as my regular posts. At face value this was a highly effective way to write content, but was the content really mine? I feel like there are some real ethical concerns about this type of content create, but I struggle with where that line is which I will go into a little bit more down below.
The part of the experiment that I was hoping to happen-did in fact happen. My initial interest was to see if anyone could tell this was AI content and to see if they would say something.
Note, let me say the next part was from a private conversation and I have permission to talk about it in this post.
My good friend @bozz DM'ed my on Discord and had a question for me. He could tell my recent posts did not sound like me and he had to ask if they were written by AI. I was so excited, so I confessed about the post and told him all about what I wanted to do with this experiment.
We only talked for a few minutes, but the conversation I had with him is the one I want to have will all of you.
- How do you feel about AI generated content?
- How did tell AI vs Human content and is that line hard to see?
- Even thought a post could be generated by AI it can create real conversation. Does that make that make the whole piece more human even though a machine made the prompt?
These questions are at the core of what I'm trying to figure out now that AI is here and innovations around it are not going to slow down. I nervous and excited about what it could mean for platform like Hive and Leofinace. I'm excited that it could help the pace at which content is created, but I'm nervous we will loose any value that actual content has.
I guess only time will tell what and how AI will be used in the future, but for now I hope we try to use it responsibly. I'd love to get everyone else's opinion on the subject.
Cheers,
Cryptic
Posted Using LeoFinance Beta
AI will truly be the destruction of humans as we know it. I know we have cell phones and instant access to all kinds of stuff but I am not looking forward to the day where people use the AI bots for everything. It’s going to completely lose what it means to be a human.
Posted Using LeoFinance Beta
I feel you I hat the AI customer service chats and calls. If this is any indication of where we are heading I actually have some legitimate fears.
This technology is good and has been doing for a long time now a days but I have to say here that it cannot do what human brain can do. Some of my friends have also used it. After some time, if he starts giving wrong answers, then it can cause a lot of damage to the person.
You bring up a great point about the loss of credibility if it gets something wrong. Just look at Googles recent launch of their AI that got a fact wrong and their stock got clobbered.
You are absolutely right, I just realized how much damage Google has suffered. Even though CHATGPT is also famous, in the coming days we will see him getting paid up too.
It is impressive to see how much AI has advanced in recent years, and even more impressive as new companies get more and more involved in it every day, what will it be like in 10, 20, 30 years? I think it is a doubt that more than one of us have.
I think it will be much soon than that. At this pace 2-5 years and I believe AI will be so good that it will be very hard to tell the difference.
I came across this post when @bozz reblogged it and checked the three posts you 'wrote'.
What I find is that AI generated content is very cold and impersonal, and quite repetitive like the Jefferson post, all three paragraphss starts with Jefferson.
In recent months I'm pretty sure there has been tonnes more of AI written content on Hive, a lot of what I call fluffy content on topics that do not fit with the author profile like a 22 year old talking about how to deal with divorce, or an author who writes in depth posts on a range of subjects from complex medical conditions, LGBT issues, gardening, mental and psychological issues, economics and finance... And when people comment on their post, the author's response is no more than 10 words if you're lucky.
AI is going to be a big issue for Hive and personally I'm going to have to support and curate people I trust only now. No more trying to support minnows as soon as they pop onto Hive. They'll have to earn my trust.
Sadly many still curate blindly instead of actually reading content. You have a good friend who read you content and can spot that something is off.
I agree with you that there seems to be more AI content as well. I see similar trends or I get an off feeling about some posts. I'm hoping that the connections we make will get stronger as we need to learn and interact more with each other as a form of verification. I do want to give people the benefit if you don't follow someone closely this content can be hard to filter with out additional content. Moving forward I know I'm going to be more carful about who and what I curate based on experiment.
AI is truly changing the way we work, but still it is lacking the emotions we can add to our post, ChatGPT can write post for you, but it can add your emotions in the post. That's the different between AI written blog or Human written blog, apart from this, ChatGPT is revolutionary.
Have you tried changing the prompt? I have seen some people to tell it to write in certain tones and it gets pretty convincing if you ask it to do the right thing.
No i haven't tried it, i would love to try it, if it also can happen, it would be amazing.
The other day I saw a person organizing his life through AI. But actually everything has a downside. Misusing it will harm us. AI-derived content likely contains plagiarism.
Yes, I think a lot of people will claim AI content is plagiarism. The odd part is if you ask AI the same prompt multiple time is produced different content so is that content consider original?
AI is a topic that is very controversial because there will be a ton of people losing their jobs over it. I think it can still be caught right now but I saw someone using the Bing AI and it was quite decent. There are ways to improve it but it could search through data and give you results easier.
Posted Using LeoFinance Beta
Bing was better than google! Googles AI said an incorrect fact during their trial which was caught by some smart people and it tanked the stock.
Cool experiment! I'm going to check out those AI posts.
I thought so I was a bit torn using it, but based on the feedback I'm getting I think it was worth it to start this conversation.
Oh, the experiment was certainly worth it, which is why I spent the last two hours looking at five of your posts in depth. My final impression is that AI generated content may be okay (once it gets over its blatant repetitiveness) but is still as dry and boring as ... a lot of classic mainstream media, actually. 😳
So is it really worth it? For cutting corners in your English class it may be, especially if your teacher is not the most tech savvy one. However, the 5 minutes it takes to organize the prompt and let the AI come up with a 200-400 word text, which then still needs to be proof read... I'm not sure if you couldn't write a genuine article in that time, which may be decent if not good, and will in fact help you develop your own skills (if that's something you actually care about).
Regarding milking the system, I'd say there is one important feature that makes a Hive post (in fact, any newspaper or magazine article) interesting, is the personality of the author. And that is something I can't see happening with AI. In this regard, I would completely forego upvoting boring posts, simply for lacking unique human authenticity. Since I can't prove that it's AI generated, I consider a downvote as too drastic.
And since I mentioned the dry and boring style that has been held up as the pillar of objective news reporting, I'd say that this innovation of AI texts is a perfect nail in the coffin of classic media. At first it should be the writers to embrace this tech, saving on their ever increasing workload. No editor is going to have a problem with the impersonal and characterless articles produced in this way, until they figure out that they can produce them completely without the writers. But until this happens, it's in the interest of every mainstream writer to use AI to milk that outdated system for all it's still worth, before it collapses in itself, while at the same time building their escape boat, for example on Hive.
Human authenticity is key and I thing AI can try to mimic it if given the right prompt, but once you make a connection with someone it's much easier to tell that their voice is gone. I like your perspective on Hive content considering I agree boring content does not get upvoted on that basis alone. The part that worries me is automatic curation trails/auto voting that can b e gamed with lots of AI content.
Yes, I can see that. But then again, who do you autovote? People that you know, and know their writing. And sure, they may pull off a couple of AI posts, but since I'm familiar with their style, it's just a matter of time before I'd notice that something is off. After all, I follow them for a reason, and while I may not read every single one of their posts, I generally am happy to open them to see what they've written.
People are getting more involved in AI generally than human and most times I usually wonder what will 2050 look like
Yes we are I just watched a few youtube video on deepfakes which are even crazier than ChatGPT. When it's hard to tell if someones face and voice are real in a video we need to figure a method to screen content.
Looking back now I can see those three, but at the time I only caught the one. I really think there are too many glaring errors out there right now. Just look at the recent Google debacle. I don't think we have achieved true machine learning yet in that the AI can only regurgitate what it finds. When it can start vetting that info, then I think we will be close. What I have found is most AI follows a specific pattern for its content. Especially ChatGPT. Once you can recognize that "formula" it is easier to realize it wasn't created by a human.
Correct the google screw up is such a perfect example of how AI can make it look good but can't distinguish between fact and what's on the internet. It also is not very creative like you pointed out. It follows a structure and is likes to reiterates parts that seem inconsequential but if you notice it the pattern is obvious. I think in a year or two it will be harder to pick up on these patterns.
Yes, I agree with you. It's easy to forget this is still in it's infancy.
Scary to think is more like it.
Congratulations @cryptictruth! You have completed the following achievement on the Hive blockchain And have been rewarded with New badge(s)
Your next target is to reach 10000 replies.
You can view your badges on your board and compare yourself to others in the Ranking
If you no longer want to receive notifications, reply to this comment with the word
STOPCheck out our last posts:
It is like Photoshop. It can alrer the whole image but still the original images have their own values and edited ones are the demand of time. Adobe and Shutterstick are working on a project to check the image and know the original one. Soon we will be able to see on their website. I think the same will be with AI generated content. There will definitely evolve the ways to check the origin. Deepfakes are also a matter of concern these days.