Creating a a semi-intelligent Curation Bot for HIVE!
Earlier today, I stumbled upon a fascinating blog post titled Refactoring My Auto-Voter Code. One line, in particular, caught my attention:
"I was trying to build a semi-intelligent Curation Bot that would do a bit of analysis on posts and then vote accordingly"
This insight from @buggedout sparked an idea, and I immediately set to work. The result? The first phase of a Semi-Intelligent Curation Bot, crafted from scratch tonight using Node.js. You can explore it here: gptvoter on GitHub
This is merely the beginning. Yet, it serves as a prime example of integrating ChatGPT with the HIVE blog blockchain to add a layer of intelligence to your scripts. My experience programming financial trading bots for ChatGPT laid the groundwork; analyzing blogs isn't all too different from dissecting stock reports, financials, and news.
How does the script work?
The script operates on a simple premise: it fetches posts and decides whether to upvote based on the calculated blogImpact score. Here's a step-by-step breakdown:
- Initial Setup: Enter your settings, fill in the .env with your keys, and secure a ChatGPT API key (fortunately, 3.5 has always been freely available for me, but keep rate limits in mind).
- Tag Configuration: Choose the tags you wish the bot to monitor (e.g., 'technology', 'pob', 'hive') and set a frequency for checks. For testing, I've set mine to every 10 minutes.
- Post Fetching and Voting: The script assesses posts on a scale of 1-100. Posts scoring above 70 receive a 100% upvote.
This iteration of the script is just a starting point. I have numerous ideas to refine and enhance its capabilities. For now, it's a solid foundation for anyone interested in blending ChatGPT's prowess with HIVE.
code redacted!
Future Enhancements
Looking ahead, I plan to enrich the script's analytical depth:
- Grammar Analysis: By integrating a Natural Language Processing (NLP) library or API, such as node-nlp or retext, the bot will evaluate posts based on grammar quality.
- Further Metrics: Other considerations will include post length, views, and more. With AI at our fingertips, the possibilities are boundless.
I'm eager to hear your thoughts, questions, and feedback. Feel encouraged to dive into the GitHub repository, clone the project, or push improvements. Your input is invaluable as we explore the frontiers of AI in content curation together.
Stay Connected
For more insights and updates, follow @coininstant. Your support and engagement fuel this journey into the future of automated curation.
@tipu curate
Upvoted 👌 (Mana: 41/61) Liquid rewards.
This is awesome idea. You should be thanked. I am surprised on that you know coding
Thanks, I surprise myself sometimes too! lol
Really enjoyed reading this! Keep up the great work.
I write posts for real people to read, not for bots to rate.
The logical continuation of your idea will be that robots will also write posts. (This is already happening, of course.) So robots will "write" and "read" and humans will be useless on Hive. I understood it correctly?
I appreciate your perspective, and it's a valid concern that the rise of bots in content platforms like Hive could seem a bit dystopian at first glance. However, let me assure you, the essence of Hive, powered by real human experiences and creativity, remains untouched at its core.
The introduction of bots, particularly those designed to curate and rate content, isn't about replacing human interaction but enhancing it. Think of these bots as your highly efficient assistants, sifting through vast amounts of content to highlight gems that might otherwise be buried under the "noise." Their role is to ensure that quality posts get the visibility they deserve, aiding in the discovery process for users who seek valuable content amidst a sea of information.
Indeed, the technological leap has brought us to a point where bots can generate posts, complete with visuals and narratives. While this sparks a debate about authenticity and creativity, it's crucial to differentiate between content created to enrich our experience and mere spam. Disclosure plays a key role here – as long as it's clear that a bot has produced a piece, readers can make informed choices about what they consume. This transparency fosters a space where AI-generated content can coexist with human-created stories, each serving different interests and purposes.
Moreover, Hive's decentralized nature and community-driven governance model provide robust mechanisms against misuse. Users and creators have the power to flag and block spam or abusive behavior, maintaining the integrity of the platform. It's a collective effort to ensure that Hive evolves in a direction that benefits all members of the community.
As for the tools provided by Hive and OpenAI, my role is simply to explore their potential to enhance our platform. It's not about sidelining humans but rather augmenting our capabilities. The vision isn't a Hive devoid of human touch; it's about leveraging technology to amplify the reach and impact of genuine, creative expression.
And let's not forget, there are irreplaceable aspects of human experiences that AI is far from replicating. The nuances of a travelogue penned by someone who's felt the sand between their toes, the emotional depth of a personal story, or the authenticity of a photograph capturing a fleeting moment – these are treasures that AI cannot mimic.
So, while AI may assist us in many ways, the heart and soul of Hive will always be its human users. We're not stepping into an era where humans become obsolete on Hive; we're entering a phase where technology and humanity converge to create a richer, more diverse platform. And that's something to look forward to.
Thank you for exhausting answer. What can be used can of course also be abused. But almost anything can be abused. After your answer, I believe that OpenAI can be beneficial for Hive.
I am pleasantly surprised by the clarity of your views on the role of artificial intelligence in the Hive. I admit, before your exhaustive explanation, I had some doubts about this, seeing the amount of controversy that had already taken place around this topic. But your text helped me take a conscious position.
AI is only a useful tool in our hands, and how to use it depends only on us
There will always be people trying to do harm, but it is stupid to ban cash just because someone is selling guns and crack for it (although we are moving in that direction). And AI is exactly the same.
I will wait for the appearance of a convenient UI, because unfortunately I don’t understand anything in the code))
Well, people write bots to make curation easier :D
Great idea to stimulate positive activity among bloggers on the hive! (I can expand on this statement deeply) It may be worth covering comments as well, even outside the tags that were initially of interest, if this does not greatly overload the operation of the equipment.
Maybe, based on artificial intelligence and your developments, it is possible to create something like hivewatchers, only more effective and tireless? I can suggest criteria by which you can find abusers if someone is seriously ready to do this
I’m probably not the only one who would support such a proposal!
Man, what you do is interesting.
Thanks, Lupega! I appreciate your interest. What excites me the most is the new-found symbiotic relationship between AI and blockchain. The transparency and security of blockchain add a layer of trust to AI operations, while AI can enhance blockchain's efficiency and user experience. This project is just a glimpse into how AI can add value to HIVE, and I'm excited to explore further possibilities. The future is bright, and it's thrilling to be part of this journey at the intersection of AI and blockchain technology. It'll be interesting to find out where this road takes us in the future!
It sounds good and all and can help good authors get rewards even if they are new and/or not "wellknown" here on hive :D
Nice work. I hope it functions as you hoped. In the end I didn't find the time to do something smart like this but I think it's a great idea to try.
The little script I started with keeps growing and getting more complicated. Eventually, I'm going to have to do some refactoring of the code myself, I'm not even there yet. I plan to start doing longer test runs on my server tonight, have a few little tweaks to do with the timing for targeting vote power levels, but for the most part, it is running exactly as expected during the tests. I have plenty more enhancement ideas, I think working on this little bot will keep me busy for a long time to come.
Aha, now i see. i'll observe with interest.
Fyi: https://salt.security/blog/security-flaws-within-chatgpt-extensions-allowed-access-to-accounts-on-third-party-websites-and-sensitive-data
Project is over, everybody (the whales) hate ai here apparently, and don't even allow ai auto responders so I shut it down after receiving too many flags on the burn comments.The ai bot was just an experiment, I was having a lot of fun programming it, I needed a live environment to test code, & hive is a very useful realm for this sort of testing. I took the open source code for this project down after i saw the reaction it received when in use. As for your Fyi, that is not anything i'd be concerned with, I don't use any 2nd party chatgpt extensions.
Thanks for info. It's good to experiment.