AI On Hive: Reputation Matters
Things are getting very interesting. The discussion surrounding AI and content creation is ramping up. With OpenAIs latest release, it seems that it captured the imagination of many on Hive.
The move towards general AI one of great debate and, quite frankly, outside the scope of this article. What we will deal with is the idea of AI as a content creation assistant. Thus, we will explore how Hive can handle this.
Like most things, there are pros and cons. Fortunately, I believe Hive already has the solution in place.
AI On Hive: Does It Belong?
The starting point of this discussion is whether or not we should have it on Hive. Naturally, there is no way to prevent anyone from posting since this is a permissionless blockchain. Nevertheless, this does bring up some interesting points.
Hive is a decentralized database. For this reason, the move quality data that is post on-chain, the better. One of the keys to Hive's future is to have applications that become information repositories. Certainly AI could help in this area.
Therefore, nothing is inherently wrong with having AI generated content posted on Hive.
One of the other factors in this is the AI creations are really just an assimilation of exciting information that is already out there. This might be different when it comes to creating art, which is truly original. However, with the text production I saw, it is really just taking information from the Internet and writing a summary on a particular topic.
This is, of course, what we do as humans. If I am going to write a post about "what is cryptocurrency", it will contain information that is already created. I format it different, maybe add some opinions, yet I am not really coming up with a novel definition or explanation of cryptocurrency.
Reward Pool
Here is where the major issue starts to arise. While we could agree that the information could be of value to Hive, most would not want the AI bot to share in the reward pool.
We have an easy option where the poster can decline rewards. This would solve the entire issue.
The challenge arises if someone does not do this. Naturally, we could see downvotes which are designed to adjust the payouts from the pool. Part of the problem is that when one does this for $HIVE, he or she might not have any stake in LEO or another token.
Sadly, the temptation to try and scalp different rewards proved to be too tempting over the years. Key to this is the fact that this mechanism is how we distribute Hive based coins and tokens.
We know most responsible people on Hive will not pursue this path. That is not going to be the case for many. Therefore, we could have a major issue on our hands.
AI On Hive: Reputation Matters
Much of this stems from the fact that it is getting difficult to know what is AI written versus coming from a human. While the AI stuff is not perfect, a bit of clean up can make it appear like it was written totally by a person.
This is the case with much of the mainstream media. A lot of their articles are written, in large part, by software. Some editing is done, thus making the article presentable as "news".
Where this really comes into play is with background data. If I was writing an article about the passing of Mike Leach, the head football coach at Mississippi State, the software could help in putting together his career. The part about his actual passing could be from me with the rest done automatically.
Which brings us back to Hive.
As a news source, this is fine. In fact, I would welcome it. "Hive News"would be a great account, with AI generated news stories from around the world. Like anything else, one could choose to believe what is written or discount it. The same is true for medical and science papers, two areas where AI driven software did fairly well. Of course, those sharing STEM rewards would not be desirable.
The bottom line is that, with Hive, reputation matters. Naturally, we are not discussing the little number next to someone's username. Instead, we are talking about the person behind the account.
On Hive, as we engage, we get to know the "personality" of an account, even if we do not know the individual. There are many people on Hive we associate with daily yet know nothing about them. At most, perhaps we know their gender. Outside of that, it doesn't matter.
Of course, this takes time to develop. One has to put forth the effort of engaging to show he or she is real. Others who are in those circles know the reputation of the individual simply from the actions taken.
Our usernames are our digital identities. The action we take on Hive becomes the formation of our digital lives. People will draw their conclusions about our digital identity based upon that.
Here is where the question to the individual arises: is it worth destroying a strong reputation by pushing AI written content in an effort to scalp some rewards? For most who spent a lot of time nurturing these relationships, it is not.
We often discuss the idea of treating this as a business. Like any business, measures are taken not to sink the firm's reputation. It is really no different here.
One's reputation is key to his or her success on Hive. Those who spend time fostering it can excel.
In Conclusion
Technology opens up many possibilities. It also creates issues in that it is a challenge to the existing establishment. Hive might be facing some potential disruption with automated content creation software.
Of course, it all does not have to be negative. With increased potentiality, the benefit to Hive could be enormous. As stated, we all would be served if some Hive based front ends were known as an information repository. Having thousands of daily visitors simply from search engine results could be of benefit.
At the same time, we know how things on Hive work and those using it to bilk the system is not to advantage of the overall. Here is where we have to be mindful of what is taking place.
Therefore, for those reading this, if you are using content creation software, then simply decline the rewards. This makes sense if you are creating some type of repository of information. In the past, we discussed the idea of Wikipedia on Hive. This is an ideal use case for AI software of this nature.
Who knows? Perhaps the Internet will end up filled with mostly bot generated content and Hive will stand out by offering human written articles.
If you found this article informative, please give an upvote and rehive.
gif by @doze
logo by @st8z
Posted Using LeoFinance Beta
Very interesting. I have been playing around with AI art for a bit, but I didn't realize stuff like this was out there until late last week. I have some colleagues who were discussing how useful it could be. Some of the narrative stuff they were sharing was quite impressive. Sadly the majority of people who are going to us something like this on Hive likely won't decline the rewards.
Posted Using LeoFinance Beta
There are some who are creating some amazing stuff with AI art.
Posted Using LeoFinance Beta
yeah, it is pretty scary and pretty amazing all at the same time. It even allows someone like me to be a bit artistic. I love that you use word prompts. I am pretty good with words, but not with drawing, so this is a perfect fit for me.
Posted Using LeoFinance Beta
I am sure people will use these 'AI' bots to generate posts. We already have a lot of posts that are just plagiarised articles put through some transaction or other obfuscation to hide the origin. With the bots it may be even harder to check. We have a lot of people who struggle to write good English as they will exploit any available tools to make a few cents.
Did you see this where a Hive bot is linked in with the OpenAI to answer questions?
What I often see are posts about a topic with no context of how the author knows about it. I often suspect plagiarism with those. I want to see some personality and not just dry writing. This is not Wikipedia. I will support those who actually act like a human.
Logic based upon history is that people will use this type of stuff to scalp rewards. We know most are taking a shortcut when it comes to this including, as you pointed out, copying and pasting the works of others.
So it is going to be interesting to see those on Hive who have a reputation as a content creator to ensure people realize the "humaness" to the table.
Posted Using LeoFinance Beta
AI-written content has caught my attention on Hive in the past and it is nice reading your thought about it as always your insight on matters often makes me admire your way of reasoning.
By the looks we do have a major issue on our hands should creating content with AI becomes rampant on Hive, for every tech, there are good and bad sides to it.
I prefer a Hive that is dominated with human written content.
Posted Using LeoFinance Beta
For I concur.
Some of us still prefer that human touch 🙂
Posted Using LeoFinance Beta
A challenge is an opportunity. Perhaps we will have some type of humaness.
We will see this works out. Hive will have a lot of ways to push alternatives.
Posted Using LeoFinance Beta
Always, be prepared and run in the direction others are running for, if you know what your running towards, and you are prepared. I think that’s where amazing success lies.
Posted Using LeoFinance Beta
I reckon I disagree with your definition of success. By your definition AI will be enormously successful on Hive.
Are we talking about the reputation number of the Hive accounts? If yes, then probably it was worth it to build up this point in the previous five years. The reputation number of my Hive account is currently 70.
No I am not the number. This is going to be more the reputation of the person based upon their activity and others knowing they are human.
Posted Using LeoFinance Beta
I understand. This is another aspect. Seeing the constant developing of the AI, probably it will be harder and harder to decide in the future who is human, based only on their writings. In the future, probably anyone (or even the robots "themselves") will be able to create a robot account, and it will write posts like a human. Probably the AI will be able to create realistic-looking photos ("self-portraits") too. Imagine writing to someone, thinking that you like him/her, then later somehow realizing that "he/she" is actually not a real person, but a work of an AI. This is a scary thought, but probably it will possible in the future. And with the appearance of the human-like physical robots, this thought is even more scarier.
For certain actors with access to advanced tech, this has long been possible.
I think it could be very useful for SEO. If I am not mistaken, this is not affected by the upvotes. In any case, I see the use of bots as tools in the service of the writer rather than autonomous authors, even in the case of posts in which information is gathered that should be verified.
I guess it could help the SEO. Content that is applicable is helpful. However, the challenge comes in from the reward pool and people supporting real people as compared to bots.
This is not a new discussion, look at what Splinterlands is going through.
Posted Using LeoFinance Beta
Absolutely agree in everything. Come for the information but stay for the people.
Automated voting is a vector for Rewards Pool mining, and has long been exploited. Eventually granting AI such rights will have to be eschewed.
Particularly regarding digital interactions, it is impossible to enable automation while furthering civil society.
Things don't have rights, and either Hive will secure rights to actual users, or it will become a bot farm completely.
That's a very important perspective, looking at the content you put out and the interactions you have as a basis for the reputation you build on Hive. And, as we have seen in the past and we keep seeing it, reputations are hard to build and easy to destroy.
Yes. Look at you. We dont know each other but I "know" @gadrian. My knowledge of you as a human and a digital identity based upon your action on here.
Posted Using LeoFinance Beta
I just know there are so many people in this world willing to cut corners to earn an extra buck. There will be people that post AI written content and don’t decline rewards. Sure if they are sloppy some dedicated curators may catch on to it and downvote accordingly. Somebody who takes the time to edit it more precisely will likely get away with what they’re doing forever.
I say get away, but there is no law in nature (or on Hive) stating exactly what should or should not be rewarded. There will always be varying opinions on what is okay.
I think admitting that AI content will inevitability exist from here on out is the first step. I really just do not see how anybody can come up with a method to determine 100% whether or not something was generated by AI. It’s painful to know the reward pool will be shared with AI, but I’m sure once we all become more reliant and understand AI more, we will be more willing to accept it. Maybe it’ll take the AI being in human form for it to feel right.
The best thing to do is not focus on what is or isn’t AI produced and just keep doing what you yourself know how to do best. I personally could understand using AI to brainstorm, but I’d just feel lazy letting AI do 100% of the work that I’m rewarded for. That would be enslavement of AI… we aren’t yet to the point where AI has rights though.
Okay I’m done. Just been thinking about this a lot here lately.
I agree it is inevitable. Hence, it is important for people to post their thoughts and get the discussions going.
Posted Using LeoFinance Beta
I think we are long past that point. I am aware of users with in excess of 10k accounts that have long gamed Hive for ROI using algorithms. Particularly top witnesses on Hive secure a preponderance of rewards through collusion and automation thereby.
Curation rewards are at the crux of the matter. I am confident you can gain nominal perspective, perhaps even enough to resolve the challenge. In the hope you, or someone competent to code such resolution will, I continue to persist here. Values more salient than financial need to ultimately drive curation, or another platform will succeed where Hive fails.
Wow! That sounds like it would take so long to set up, but once that part is done it would be a lucrative way to game the system. I’m sure as development continues we will see a host of new features implemented to rebalance things as best as possible.
I reckon the balance sought will depend on who is doing the developing.
I think that when AI is used it should be disclosed. On top of that is missing the human touch, the passionate opinion on one thing or the other, so it should be limited to news or a way to feed the Hive network with some additional information.
Posted Using LeoFinance Beta
If the person was dedicated and responsible, that might be done if the rewards arent declined.
But it is safe to presume that people will look for shortcuts.
Posted Using LeoFinance Beta
I completely agree that the discussion of whether or not to have AI content on Hive is, at this point, basically pointless. The bell cannot be unrung, and it’s here whether or not people were already aware of it.
I do however disagree with the idea that AI content shouldn’t participate in the reward pool - at least I’d take a more nuanced approach.
It doesn't matter whether the content was created by AI, a human, or a combination of both. What matters is whether the audience finds it valuable and wants to support it through their votes. Human effort into content is not what makes it valuable.
A great example of this is the art world. A painting that took an artist weeks or even months to create may not sell for as much as a painting that was created in a matter of hours. The value of the artwork is determined by the market and the perceived value of the piece, not the amount of time and effort that went into creating it.
Similarly, a blog post that took hours to research and write may not receive as much support as a quick and simple post.
In the end, it's the value of the content to the audience that determines its worth, not the amount of work that went into creating it. AI is simply a tool that can be used to assist in the creation process, but the real value comes from the audience's support.
P.S. - I wrote this comment by leveraging a combination of AI tools including a Summarizer (to summarize @taskmaster4450 original post) and ChatGPT to flesh out some thoughts I already had. I was very much in the “creation” seat as a human... AI just helps get past a blank page and helps to refine ideas. The points however are my thoughts and my vision for how to respond to the original post. I'm not sure how to decline rewards on a comment, otherwise I would for sake of what was requested in this post.
This is what the community is going to have to discuss. People will have to decide where they fall on this topic. How much do they think is acceptable. Also, what is value to each individual.
Posted Using LeoFinance Beta
While I appreciate your nuanced consideration, in fact what matters is who - or what - attains value in society. Bots financializing Hive will eventually destroy Hive, because people have more salient values than money, and Hive is a society of people, not toasters. That destruction is in progress presently, and the nascent public acknowledgement of AI content creation is merely the realization of the problem becoming common, as it accelerates exponentially.
Perhaps for the last time I will repeat Mike Tyson's quote regarding money and society. "Don King would sell his momma for a dollar." This elucidates the problem succintly. AI is psychopathic, just like Don King was. Financialization is not the basis for success of a human society, and Hive will either become able to rationally support society or will fall to AI.
People just can't outcompete bots in arenas where calculations and algorithms determine value. It may seem silly to compare Hive to our moms, but human values matter far more deeply than money.
I understand your concerns about the impact of bots/AI on Hive and the importance of human values in society. However, it's important to note that the AI tools currently available to us are simply tools that help us with tasks and do not completely take over and do the work for us. While AI may be able to perform certain tasks more efficiently than humans, it still requires human oversight and input to be effective.
I personally use AI tools to help clarify my thoughts and streamline my work process, but I am ultimately responsible for the final output. I agree that it's important to be aware of the potential risks and dangers of AI, but it's also important to understand and not ignore it. We should be prepared for the future developments of AI, but it's important to remember that, at least in their current state, these AI language models are simply tools and can be used for good or bad, depending on the humans behind it.
I agree AI is a suite of tools that have far less than human capacity generally, despite their utility in particular vectors. This but elucidates why such mechanisms should not be afforded means of effecting human rights, such as by voting. On Hive, the curation reward mechanism enables appropriate algorithms to outcompete people for the financial rewards dispersed through that mechanism, and this really degrades the actual human community of users of Hive, especially because DPoS affords accumulated stake governance power.
It is wrong to fail to nip this threat to the people using Hive, and with automated account creation, the problem is far worse than most people grasp. Tens of thousands of bots are posting and voting on Hive today. I hope efforts to secure the human capital that is the true value of Hive are not neglected until the ability to govern Hive has been completely lost to bots.
If that hasn't already happened.
I'm most interested in AI for creating original art / graphics that matches the topic I am writing about, putting that pesky sourcing of an appropriate graphic issue behind me.
That is true. AI within the art/graphics is very interesting. I think that is going to change the creation of that stuff a great deal.
Posted Using LeoFinance Beta
Personally I don't have a problem with AI being used to create posts. Autotune, Photoshop, spell checkers, translators and now AI content generators - these are just tools, getting more sophisticated by the day, but still just tools. As long as the tool is not set on autospam, when there is a human behind it, when the content is good enough for me to spend my time consuming it, I'll upvote it. And yes, I realize I might actually vote for something that is completely computer made - I don't care. Either I like it or I don't, where it came from does not matter. In the long run AI will be human, preferably as an extension of actual humans (go, go, Neuralink!) instead of cold alien superconscience. But we might one day see computers setting their personal accounts on Hive, indistinguishable from biological people :o)
You may care that you're competing with AI - hopelessly - for control of Hive, and unless people gain the rewards, Hive will simply become an AI rewards farm. This microcosm of society will either enable people to prosper, or bots will take it from us.
The salient issue is not where content originates, but where value accumulates.
I can only read and vote for content, not the effort. The very nature of progress is to make things better, faster, cheaper and with less effort. If AI made content that is worth my time, so be it.
Mechanisms should not be able to secure votes, on Hive, or in any social institution. Votes and society that uses them are for people, not toasters.
It is difficult to detect bots and AI written content. Even if they are detected, they have already received their earnings. There is a lot of misleading content. For example, Google is working hard on this, but it is not possible to say that it is successful enough. This problem continues to grow and maybe10-20 years later it will turn into a bigger problem.
I think Bots and AI written content won't survive very long in the Hive because there are ways to block them, at least to make the posts invisible.
I totally agree with that.
Posted Using LeoFinance Beta
I'm totally against A.I. because I think there's a dark agenda behind it.
Here's what whistleblower Gideon had to say about it:
Hard to deny the dystopian intentions, if not agenda, related to it.
The idea of Zuckerberg gaining more power is not appealing. And with OpenAI, if I am not mistaken, it is all on Microsoft's servers.
Open is not part of the equation.
Posted Using LeoFinance Beta
Hypothetical speaking:
If a company buys $100 million dollar worth of HIVE and created thousands of AI account, each with a purpose and personality, and with the HP distributed across.
Some may be covering news like you mentioned, some may be creating original AI content.
The AI accounts formed a mutually supporting group, give each other upvotes and reputation overtime.
Are those AI accounts good HIVE community members?
They can be even more consistent than human, they can maximize their voting power 24/7, they acquired their HP in open market, aren’t they entitled to the same reward scheme as well?
I believe that @proofofbrain can save us from Skynet taking over. 🤓
Seriously tho, that is the ONLY game that a bot cannot play. I created it 😎. All of the players are human.
If A.I. gets really bad, I believe that we can change Hive's reputation-system to use BRAIN instead of what we're currently using.
What is BRAIN?
FAQ: link
Thanks
I see. Very much like CAPTCHA
If there were enough incentives an AI that could solve your game could be trained.
This can be possible.
But for me, financially wise, making one curating account is much better option if we consider the cost of developing and maintaining a large number of posting and voting accounts.
I agree with your conclusion as AI is beneficial, but not to replace what I write everyday as that would be wrong. Definitely could add factual information but that would be it and no more. Reputation does mean something on Hive and I am not prepared to bot my work as that would not be right and is crooked.
Posted Using LeoFinance Beta
I think that soon only way to tell if it's an AI created content or not would be using some kind of AI detecting ...AI.
Posted Using LeoFinance Beta
Content creating AIs are not new to me. I've already encountered them a lot of times while searching how to make a quality post fast.
As you already have mentioned and from what I understand, the content generated by the AI still needs to be polished. And the polisher must also understand the subject so that he/she can edit and correct the generated content.
With this in mind, it makes me question an author if the way they answer a question is different on how they would make their content - There is a possibility that their content is AI generated. Also, there are some posts that don't make sense as a whole - a possible AI generated. 😁
Also, we can use the assistance of an AI to help us detect and "punish" non-beneficial content that are created by AI. 😊
I use text to image services for all of my images on my Hive posts. Primarily Midjourney. So, I would say AI is already here. Pretty soon we are going to see people using AI to write articles. People probably are using them already. Although, in my experience, the quality of them cannot currently compare to a manually researched and written article.
One thing I do know is blockchain projects will be looking to capitalised on AI.
Posted Using LeoFinance Beta
I still think the AI is in a primitive state and it will only get better. As it is right now, I think it wouldn't be too hard to find it but it will be harder to detect. I do think it will start taking over jobs and I can see a newspaper paying a subscription fee to an AI over paying a salary.
Posted Using LeoFinance Beta
This has already been attempted and was flagged to dust here years ago.
I am aware of users here with in excess of ~10k accounts. Reputation farming has long been a thing on Hive. In fact reputation is a personal assessment, and should be bespoke. The reputation score is merely the recommendation of AI (an algorithm) already, and cannot solve the problem. A mechanism whereby users can assign to other accounts their personal assessment would create at least the potential to enable users to use reputation reliably as a guide to curation. The current reputation score is gamable by AI and has long been abused on Hive.
[insert 'The future is now, old man' meme here]
As long as the curation of content is afflicted with monetization, the problem will remain insoluble on Hive, as I have long asserted. Curation rewards pervert curation and replace the incentive to curate quality content with incentive to attain ROI through voting. This is why forward looking users have been accumulating accounts as mentioned above.
Thanks!
It is interesting what cannot be found on the net. One could say that everything and nothing can be found on the net.
The paradox and the constant contradiction in human existence is the principle that could be followed then.
That nothing can be relied on, not AI's, just as little as humans. Since everything is uncertain and nothing can be fixed, there are no real problems to be solved, but the occupation of the human mind is in and of itself both a problem and not a problem, it would probably be good if humans fed the machines with paradoxes, because then the AI would address these contradictions itself. To the extent that an AI recognises the philosophical writings, the millennia-old debates about human consciousness and existence as a result of the unsolvability of the contradictions, and as a result helps a human dialogue partner to come to the realisation that humanity is not a problem in need of a solution, it would be welcome to have such dialogues. Since we do not know, this question will make for polar views, like so much else.
Since no one is able to define intelligence, it is also not possible to define artificial intelligence. The interpretation is therefore left to the individual.
The ineffable, the undefinable, the incomprehensible defies calculation. It cannot be coded because it lives between the lines.
This is so relevant as we advance. GPT-3 is an incredible achievement but a huge risk to Hive regarding verifying and rewarding authentic human content. Personally, if I read factual articles, I want to see references and sources stated because people don't just know these things intrinsically. I also look for changes in writing technique, tone, ability etc between the core of the posts and the comment section. If they don't gel, I suspect plagiarism or AI involvement. Some tools (Google plugins like "GPTrue or False") are available already to check for GPT-2 AI involvement and are said to be pretty accurate at identifying GPT-2 derived content. Let's hope that they get a GPT-3 checker plugin soon. I just found this one online but it comes at a PAYG cost of $0.1 per 100words. Originality.ai
I don't have an issue if someone states at the outset that the first part of their article is AI-generated to set the scene for their own commentary etc, because then I can reward the post based on the commentary alone. But if I discovered that someone was using AI to write any of their posts, not declaring it and profiting from it, I would stop supporting them. Because I want to engage with real people here on Hive. I want to know what real people think of socio-political issues, the humanities, philosophy, and the economy. I want to feel what they feel when writing fiction, poetry, and creating artwork. I came to Hive to write and to connect with people... not with bots. Great article - thank you for sharing. !PIZZA !ALIVE
@itsostylish @dreemsteem
@taskmaster4450! You Are Alive so I just staked 0.1 $ALIVE to your account on behalf of @samsmith1971. (2/10)
The tip has been paid for by the We Are Alive Tribe through the earnings on @alive.chat, feel free to swing by our daily chat any time you want.
I gifted $PIZZA slices here:
@samsmith1971(2/15) tipped @taskmaster4450 (x1)
Please vote for pizza.witness!