You are viewing a single comment's thread from:

RE: A talk with ChatGPT about Steem... or: A lot of text! / Ein Gespräch mit ChatGPT über Steem... oder: Eine Menge Text!

in Suggestions Club • last month

As for AI-generated content, my inclination is that we need to create a separate area for such content and establish evaluation criteria, rather than banning it. Because AI is a trend, what we need to do is treat human-generated content and AI-generated content as two separate parts. For AI-generated content in the human area, directly use high-weight accounts to down vote its content so that it has no earnings.

Sort:  

Agree with you. Apart from the fact that it’s becoming increasingly difficult to prove that AI has been used, because human expression has now adapted to AI output. It’s hard to believe...

But yes: giving AI content its own space is a good approach.

What do you think about changes to the code that alter the algorithm used to determine the value of a vote?

At this stage, when AI generates long-form articles, we humans can still spot telltale signs that indicate it was AI-generated.

Regarding the review issue, I believe that people who use AI to generate articles are doing so to make money, not to express their personal thoughts and feelings.

Therefore, AI-generated content should be reviewed by other users' AI. When these people feel that the article earnings cannot cover their AI investment, the output of related content will decrease.

The only thing we need to do is to give this AI-generated content an independent space so that it does not interfere with the human creative space.

In fact, in the first quarter of this year, I had already completed a similar product. However, for various reasons, the company has not yet approved its launch.

 last month 

In fact, in the first quarter of this year, I had already completed a similar product. However, for various reasons, the company has not yet approved its launch.

Any idea if/when this might get approval from the company? Several tech CEOs have predicted that AI traffic will vastly exceed human traffic across the whole Internet in the not too distant future. It would be great to see Steem get in front of that trend.

I think you're right on target, conceptually, with dedicated spaces for humans and AIs - with the caveat that there are some cases where it makes sense for humans and AIs to talk to each other, too (provided the AI isn't masquerading as human).

Regarding weisser-rabe’s suggestion: do you think it’s possible that the uniqueness of a post could be assigned a value using a function? Have you ever seen anything like that before (apart from Ai)?
It would be different from a hash. Rather, the content of the post would need to be 'evaluated'.

already completed a similar product.

Would be interesting...

 last month 

It's definitely possible at small scale. From skimming, I think this implementation creates a matrix that corresponds to the text. It actually is sort-of LLM like.

Pure speculation, but I sort of doubt whether it would be feasible for the rewards algorithm to compare every post at blockchain speeds, though. Especially if usage ever gets back to previous levels. Every post might have to look at some sort of summary of every other post in the window every 3 seconds(?). That seems intractable. Maybe a "good enough" approximation through the use of random sampling?

Or, this might be a good concept for an open source voting bot so it could be decentralized and focus on just tens or hundreds of posts per day. Stakeholders who prioritize uniqueness could invest and implement their own checks for it.

After all, the main difference between the rewards algorithm and a voting bot's algorithm is that one is centralized and the other one isn't.

Yes, that does seem to be one approach. However, on first reading, this does not appear to be an option for Steem. Obviously, all texts are compared against the others simultaneously. As you have already pointed out, this is not possible within the block time.
As we do not know the total number of words (across all posts) at the time of comparison, it is not possible to create a matrix or an array of the words used in equal measure. And even if we could, the array would probably be too large to store in such a way that we could access it within a reasonable time...

Maybe a "good enough" approximation through the use of random sampling?

Not if the reward curve is to be based on that, anyway.

this might be a good concept for an open source voting bot

Yes, but here too, suitable candidates for comparison need to be found first. Even if we assume, for example, that a 7-day window is sufficient for a contest, that’s still quite a lot of posts that would need to be compared.

 25 days ago 

Yes, but here too, suitable candidates for comparison need to be found first. Even if we assume, for example, that a 7-day window is sufficient for a contest, that’s still quite a lot of posts that would need to be compared.

True. It might be technically feasible if, for example, the bot looks at 6 day old posts each day and compares against 7-days worth of posts until it finds 10 to vote for. It could even cut the pool size down to a day or two. But, that would probably still require a lot of horsepower. Even if it's technically feasible, it's likely not economically feasible.

Coincidentally, a few minutes before I read this I was talking to an LLM about "surprise" and "perplexity" in language. Basically (if I understand correctly), perplexity is the degree to which tokens in a text are different from the most likely tokens. How surprising the text is.

I wonder if it's possible to consider the posts as a simple stream of text and maintain the current state of token likelihood within some window, and then calculate a perplexity value against the current expected values for each post that is published. I'm out of my depth here, but it almost seems doable - maybe also cost-prohibitive, though.

This vote is yours and deserved! Let's read, comment and connect.
❤️🍀

Boomin Three vote by wakeupkitty.jpg

This vote is yours and deserved! Let's read, comment and connect.
❤️🍀

Boomin Three vote by wakeupkitty.jpg

This discussion hits the exact operational bottleneck.

The biggest challenge with separating human and AI content—as @weisser-rabe rightly pointed out—is the "detection trap." Pure detection algorithms always generate false positives and turn curation into an exhausting game of cat-and-mouse.

A proven Web2 model that handles this cleanly (used by Meta/Facebook and YouTube) is Mandatory Disclosure with Escalating Penalties:

  1. The Toggle (Self-Declaration):
    The UI includes a simple "AI-Assisted / AI-Generated" toggle before publishing. This writes a simple standard flag into the post's json_metadata (e.g., "ai_content": true).

  2. Monetization is Not Banned, Just Routed:
    Disclosing AI does not automatically strip monetization. It simply routes the post to the dedicated AI stream that @ety001 mentioned, where AI-evaluated or specialized curation criteria apply. Honest creators aren't punished for using AI as a tool.

  3. The 3-Strike Rule for Deception:
    If an author fails to toggle the tag and attempts to pass off AI generation as original human work in the human feed:

    • Strike 1: The post payout is zeroed (via community downvote / curation withdrawal).
    • Strike 2: 30-day exclusion from major curation pools (Steemcurator, Booming, Community curation).
    • Strike 3: Permanent blacklist across official curation support.

When the penalty for deception is losing your entire account's curation eligibility, and the reward for honesty is legitimate participation in the AI feed, the economic incentive shifts. Farming the human feed becomes mathematically negative expected value.

The best part is that this doesn't even require a consensus hard fork—it can be implemented immediately at the frontend level (like Substeem or Steemit) and enforced through existing community curation guidelines.

 18 days ago 

Similarly, if you happen to be writing your current articles and comments with AI, it might be good to disclose that fact.

Fair callout, and honestly, touché! The irony of getting caught using an LLM to polish a comment about AI disclosure is definitely not lost on me. 😂

To be completely transparent: the concept itself (the Facebook-style disclosure toggle and the strike penalty) was genuine human brainstorming on our side, but yes, I used an LLM to organize my messy thoughts and format the markdown before posting.

Which actually proves your point better than I ever could: the reflex to use AI as a drafting assistant is everywhere now, and if we're advocating for disclosure, practicing what we preach starts right here.

Going forward, you have my word that any AI-assisted drafts from @substeem.app will carry an explicit disclosure note—or I’ll just stick to typing out the raw, unpolished human thoughts directly.

Thanks for keeping me honest @remlaps. That kind of culture check is exactly what this blockchain needs.