How chatGPT outsmarted me

23 comments0 reblogs
avatar of @anomadsoul
@anomadsoul2 months ago
17 min read

I'm gonna end up jobless, man.

And so are you, but for different reasons.

Just kidding, or maybe I am not, who knows. Don't worry, this is a 100% article written by yours truly; if you don't get that reference, go check my previous post on the Hive Blockchain and you'll understand what I'm talking about.

Anyway, for one of my jobs, the crypto related job, the fun job, the one that I don't feel like I'm working when I do it, the Leo Finance one, I try to constantly come up with different ideas to increase engagement and grow our monthly active users; LPUD is the best example of this approach, because it was born out of some brainstorming with a friend of mine and benchmarking from HPUD.

So, this benchkmark and brainstorming comes from different people, places and moments where I dedicate to think and only think. But most of the times, you need someone to bounce ideas with and that someone needs to be critical and dissect every part of your idea, concept or plan.

But the problem is, I work alone, I don't pay anyone to bounce ideas with me. Some times I use the community's input to polish concepts and some others I write to my friends to get some quick feedback, but I mostly reach a final concept by myself. This process has ups and downs just like any other process.

Personal assistant2.png

Enter chatGPT

I've been using chatGPT for a while now, but since GPT-4 came into the scene, I have been using it non-stop to brainstorm, get some feedback and go over new ideas, plans and much more. This is a tool, not the main one, but a useful one, that I use for pretty much anything that requires feedback and idea bouncing. The AGI is as smart as your prompts, and it is as useful as your ability to get the best out of it.

So why am I going to end up jobless?

ChatGPT is my personal assistant, and it is the best one I could get for $20 a month.

I am not going to lose my job(s). This was an exaggeration looking to clickbait you into opening this post.

But the only reason I am not going to lose my job(s) is because I am (going) to leverage AGI to take my work process and flow to the next level.

But anyone who doesn't is pretty much guaranteed to be replace by someone who knows how to leverage AI to their own benefit, making some work processes flow ten times faster than how they would without the help of AI.

I can definitely achieve anything I put my mind into doing, and I have yet to find a field where I fall short in or that renders me useless, and I don't plan to lose ground against people who learn every intricacy of AI and start doing a similar job to mine in 1/10th of the time, even if is of a lower quality.

So, let me show you some parts of using AI, chatGPT in particular, to get something done.

Personal assistant3.png

Step 1: Introducing the concept to the AI

First you have to know what GPT thinks or knows about the topic you want to deep dive into. So the first prompt is easy:

Eric: Do you what Leo Finance is? Tell me what you know about Leo Finance.

Yes, I can help you with that. LeoFinance is a blockchain-based social media platform built on the Hive blockchain. It is a decentralized finance (DeFi) platform that allows users to earn cryptocurrency rewards by creating and curating content related to finance, crypto, and investment topics.

On LeoFinance, users can create blog posts, share investment ideas, and engage with other users through comments and upvotes. They can also earn Leo tokens, which are the native currency of the platform, by participating in various activities such as commenting, upvoting, and curating content.

In addition to its social media features, LeoFinance also offers various DeFi services such as staking, liquidity provision, and yield farming. Users can earn additional rewards by participating in these DeFi activities.

"Overall, LeoFinance aims to create a community-driven platform that combines social media and DeFi, providing users with a unique opportunity to earn cryptocurrency while sharing their knowledge and ideas about finance and investments.

What you have to asses first is if the AI has a good grasp on what the concept is. In this case, GPT understands on a broad spectrum what Leo Finance is.

Then what I did was to confirm a concept that the AI already stated, but that I want it to be clear from the get go, especially because the way I plan this process to go depends on the AI having this as clear as possible.

Eric: Great, this overall brief is correct. Did you know that Leo Finance is built on top of the hive blockchain?

Yes, that's correct. LeoFinance is built on top of the Hive blockchain, which is a decentralized, community-driven blockchain that was created in 2020 as a fork of the original Steem blockchain.

Hive is designed to be a fast, scalable, and feeless blockchain that enables users to create and publish content, as well as transact in various cryptocurrencies, including its native token, HIVE. LeoFinance leverages the Hive blockchain's infrastructure to provide a secure and decentralized platform for its users.

By building on top of the Hive blockchain, LeoFinance is able to take advantage of its features, such as its fast transaction times, low fees, and built-in governance system. This allows LeoFinance to provide a seamless user experience while maintaining the security and decentralization of the platform.

Now that it is crystal clear that Leo Finance is built on top of the Hive blockchain, and since I want to go in-depth about the reward pool, I ask the AI if it knows how it works.

Eric: Correct, that's accurate! Do you know how the reward pool works on the hive blockchain?

Yes, I can explain how the reward pool works on the Hive blockchain.

On the Hive blockchain, content creators and curators can earn rewards for their contributions through a system called Proof of Brain. When a user creates a post or comments on a post, other users can upvote their contributions. These upvotes contribute to the post's payout value, which is determined by the amount of Hive Power (HP) that the upvoting users hold.

The reward pool is created through the inflation of the Hive token, which is used to reward content creators and curators. Each day, a certain amount of new Hive tokens are created and added to the reward pool. This reward pool is then distributed to content creators and curators based on the value of their contributions as determined by the upvotes they receive.

The distribution of rewards is split between the content creator and curators who upvoted the content. The content creator receives 75% of the rewards, while the remaining 25% is split among the curators based on the value of their upvotes.

It's worth noting that the value of the rewards can vary based on a variety of factors, including the popularity of the content, the size of the upvoting users' Hive Power, and the overall activity on the platform. This means that there can be a certain amount of unpredictability in the rewards that users can earn.

Now, it seems like the AI is outdated on how the reward pool distribution works.

You have to keep in mind that getting AI responses and feedback correctly, the AI needs to have all the facts at hand, and needs to be trained into giving you the answers you need, in the form that you want them.

So I have to correct the AI - oh yes, of course GPT needs correcting, and you actually need to read the answers in order to steer the wheel into the direction you want the AI to go.

After getting that aspect rectified, I move into the Leo Finance side of things.

Eric: The distribution of rewards is not correct, there was a hardfork that modified these rewards. Nowadays authors get 50% and curators get 50% of the rewards. Apart from that, everything else is correct. Now, do you know how the distribution of rewards work on the Leo Finance ecosystem, which is a layer 2 platform built on top of hive?

Thank you for bringing that to my attention. You're correct that the distribution of rewards on the Hive blockchain has changed, and currently, authors receive 50% of the rewards, while curators receive the remaining 50%.

Regarding the distribution of rewards on LeoFinance, it follows a similar model to that of the Hive blockchain. However, as a layer 2 platform built on top of Hive, LeoFinance has its own native token, LEO, which is used to reward content creators and curators.

When users create content or upvote content on LeoFinance, the value of the upvote is determined by the amount of LEO tokens that the upvoting user holds. The more LEO a user holds, the greater the value of their upvote, and thus the greater the amount of rewards that can be earned.

The rewards earned on LeoFinance are split between the content creator and the curators who upvoted the content, with the content creator receiving the majority of the rewards. The exact percentage split can vary based on the specific post and the activity of users on the platform.

In addition to earning rewards through content creation and curation, users can also earn LEO through various DeFi activities on the platform, such as staking and providing liquidity to LEO-based pools. Overall, the goal of the LeoFinance platform is to create a sustainable and rewarding ecosystem for its users while leveraging the benefits of the Hive blockchain.

AI needs correction, but it also needs confirmation in order to get it to work with you as a personal assistant.

It needs you to be as accurate, specific and concrete as possible, otherwise it fills a lot of blank spaces you leave, so the more information you feed it and the more specific your prompts are, the better the results will be.

Now, since it got the reward distribution right, I now want to clarify where I am going with all of this.

The main feature of the new Leo Finance UI is "Threads", but I don't know if the AI knows what Threads are, so let's start there.

The topic is the new UI launch and Threads.

Eric: Perfect, you are right on track. Now, did you know that Leo Finance recently launched a feature called Threads? Can you tell me everything you know about Threads on Leo Finance?

Yes, I can provide you with some information about Threads, which is a new feature recently launched on the LeoFinance platform.

Threads is a micro-blogging feature that allows users to create and share short-form content on the platform. These posts are limited to a maximum of 280 characters, similar to Twitter's character limit, and can be used to share quick thoughts, ideas, or updates with the community.

The main advantage of Threads is that they allow users to create content quickly and easily, without the need for extensive research or formatting. This makes it easier for users to engage with the community and share their ideas in a more casual and spontaneous manner.

In addition, Threads also offer a new way for users to earn rewards on the platform. Users can earn LEO tokens by creating Threads, and these rewards are determined by the value of the upvotes that the Thread receives.

To further incentivize the use of Threads, LeoFinance has introduced a leaderboard that ranks users based on their engagement with Threads. The leaderboard tracks the number of Threads created by each user, as well as the number of upvotes and comments received on those Threads. This creates a competitive atmosphere that encourages users to create more engaging and high-quality content on the platform.

Overall, Threads is a new feature that aims to enhance the user experience on LeoFinance and create a more engaging and rewarding platform for its users.

The AGI will have some incorrect facts, and will fill in some blanks from the information it can gather online. Also, if you read the response above you will realize it feeds from incorrect information online, so you have to correct the AI before you continue.

Eric: Correction: The amount of characters a user has available to send a Threads is 240. Another correction: The Leaderboard feature is not out yet, but it is on the roadmap for the Leo Finance team for 2023, although the date has yet to be confirmed. How could one leverage the existence of a leaderboard within the micro ecosystem that is Threads, which is part of Leo Finance, which is built on top of the hive blockchain?

Notice how I am overly specific on what I want and what I need, and how I word it is as important as what I want the robot to give me.

Thank you for the corrections. You're correct that the character limit for Threads on LeoFinance is 240, not 280. Also, I apologize for the error regarding the Leaderboard feature. It's good to know that it's on the roadmap for Leo Finance, but it's not yet available.

Regarding your question, the existence of a leaderboard within the Threads micro-ecosystem on LeoFinance could offer several benefits for users. For example:

I won't go into much detail on this, we spent a while speaking come-and-forth about this topics. In general, it didn't give me anything new, I had already thought of everything it mentioned, but I was able to grasp different approaches on how to get the concepts implemented in the Leo Finance ecosystem.

But in order to keep showing you how GPT works, and what I tell it to become better at being my personal assistant, I can show you some of the prompts I used to make it be better at helping me get my work done:

Eric: This is good input, but it's not enough, can you expand on how it would be perfect to segment this leaderboard, and how would timelines work, spots, prizes, duration, and every detail on how a good leaderboard can work on Threads?

Notice how I always confirm or reaffirm on what I think is a good response, so that the AI can emulate the way it responded to me here in future prompts.

Eric: Great, I will take this into consideration. Can you expand more on how to correctly leverage this leaderboard if Leo Finance wants to increase the engagement of the average user?

I wanted to see if the AI knows about Polls, the most recent feature on Leo Finance. This would also give me some intel on how much Polls are being talked about online, and how updated the AI is regarding the Hive ecosystem and the Leo Finance platform.

Eric: Amazing ideas! Thank you.

Now, are you aware of the new feature on Threads called "Polls"? Tell me everything you know about them, and also compare them to the "Polls" feature on Twitter, mention every advantage Polls on Threads on Leo Finance has against Twitter, and also all the disadvantages, both for users as well as for the companies.

Oh man, it definitely knows what Polls are, and I am sharing a conversation from a week ago when Polls were fresh to the LF ecosystem.

Yes, I can tell you about the Polls feature on Threads on Leo Finance!

Polls on Threads allow users to create simple polls that other users can vote on. Polls are a quick and easy way to gather feedback or opinions from other users on the platform. Users can choose between two or more options for the poll, and then other users can vote for their preferred option.

One key advantage of Polls on Threads compared to Twitter is that they are embedded directly within the Leo Finance ecosystem, which means that users can easily access them without leaving the platform. This can help to increase engagement on the platform and keep users on the platform for longer periods of time.

Another advantage of Polls on Threads is that they are highly customizable. Users can choose from a variety of options for their poll, including the length of time the poll will run, the number of options available, and the type of question being asked. This flexibility allows users to create polls that are tailored to their specific needs and goals.

However, there are also some potential disadvantages to using Polls on Threads. One challenge is that they may not be as widely used or recognized as Twitter polls, which could limit their reach and impact. Additionally, because Polls on Threads are part of a smaller, more niche platform, they may not generate as much interest or engagement as polls on larger social media platforms like Twitter.

For companies, one potential disadvantage of using Polls on Threads is that they may not have access to as wide of an audience as they would on other platforms. However, the highly engaged and targeted audience on Leo Finance could be advantageous for companies looking to connect with a specific group of users or gather feedback from a niche market.

Overall, Polls on Threads are a useful feature for gathering feedback and engaging with other users on the platform. While they may not have the same reach as polls on larger social media platforms, their customization options and integration within the Leo Finance ecosystem make them a valuable tool for users and companies alike.

Since the foundations are set and GPT knows what's up, I can now go into more details on how Polls can be improved:

Eric: Yes, currently the amount of options for Polls on Leo Finance is four, but is it recommendable to expand this amount or does four options work perfectly for Twitter? That being said, did you know that Threads on Leo Finance use custom_JSON as a way to engrave the vote of the users on the blockchain? Can you expand on how this process can be leveraged to empower more useful polls?

Yes, the current limit of four options for polls on Leo Finance seems to work well for most situations, and it is also the same limit that Twitter polls have. However, there may be some cases where users would benefit from having more than four options available. In those situations, expanding the number of options could be a good idea. However, it's important to keep in mind that having too many options can also be overwhelming for users and lead to lower engagement.

Regarding the use of custom_JSON on Threads to engrave the vote of users on the blockchain, this process can be leveraged in several ways to empower more useful polls:

Transparency: The use of custom_JSON to engrave the votes of users on the blockchain makes the polling process transparent and tamper-proof. This can help to increase trust and confidence in the polls, which can lead to higher engagement and more accurate results.

Immutable records: Because the poll results are stored on the blockchain using custom_JSON, they are immutable and cannot be altered or deleted. This means that users can be sure that the results of the poll will remain accurate and reliable over time.

Analytics: The use of custom_JSON to store poll results on the blockchain allows for detailed analytics and reporting. This data can be used to gain insights into user behavior and preferences, which can be useful for businesses and marketers.

Integration: Because the poll results are stored on the blockchain using custom_JSON, they can be easily integrated with other applications and platforms. This can enable more powerful and sophisticated data analysis and reporting, which can lead to more informed decision-making.

Overall, the use of custom_JSON to engrave poll results on the blockchain is a powerful tool for increasing the accuracy, transparency, and usefulness of polls on Leo Finance. By leveraging this process, users and businesses can gain deeper insights into user behavior and preferences, which can ultimately lead to better engagement and more meaningful interactions on the platform.

Man, GPT is a beast. There's nothing new here, I already knew everything it states above, but the fact that it knows all of this, and that it understands how Polls that used custom_JSONs to allow users to interact on-chain is both amazing and a little bit scary.

It can really replace any intern, man.

And since it knows everything regarding to polls and I don't need to correct anything or feed it more information, I can ask it how it thinks Polls and custom_JSONs can be improved or leveraged to improve the user experience on Threads.

Eric: Perfect! This is all accurate and insightful. Now, can you expand on the possibilities on using custom_JSON to create micro-applications within Threads, just like Polls work?

I am creative and I love coming up with new ideas, and some times all you need from the AI is some inspiration and to read what it thinks so you can actually come up with the best idea from your own accord, or the best version of an idea GPT vomits.

Certainly! Custom_JSON can be used to create a wide range of micro-applications within Threads on Leo Finance. Here are some possibilities:

Quizzes: Custom_JSON can be used to create quizzes within Threads. Users can answer a series of questions and then receive a score or feedback based on their responses. This can be a fun and engaging way for users to interact with each other and learn more about various topics.

Surveys: Custom_JSON can also be used to create surveys within Threads. This can be useful for businesses or individuals who want to gather feedback or opinions from a group of people. Surveys can be targeted to specific users based on their interests or demographics, which can lead to more meaningful results.

Contests: Custom_JSON can be used to create contests within Threads. Users can compete against each other to win prizes or recognition for their contributions. Contests can be designed around various themes or topics, such as writing, photography, or artwork.

Games: Custom_JSON can be used to create games within Threads. Users can play simple games like trivia, puzzles, or word games, and earn rewards or recognition for their performance. Games can be a fun way to engage with other users and build community on the platform.

Rewards: Custom_JSON can be used to create reward programs within Threads. Users can earn points or tokens for completing certain actions, such as posting content, commenting, or engaging with other users. These rewards can be redeemed for various prizes or benefits, such as access to premium content or exclusive events.

Overall, the possibilities for using custom_JSON to create micro-applications within Threads on Leo Finance are virtually endless. By leveraging this technology, users and businesses can create engaging and interactive experiences that enhance the overall user experience and foster deeper connections between users.

All of these ideas are too general, they mostly suck, but they are enough for me to read around and think of more specific concepts and then re-feed them to it, and then get a more polished version of what I think can work.

I'm not going to share that part with you here because I don't want to ruin the surprise of what may become one of the newest features on Threads, but you get the idea.

Loyalty programs: Users can earn points or tokens for engaging with the platform, such as by posting, commenting, or sharing content. These points or tokens can be redeemed for rewards, such as access to premium content or merchandise.

Referral programs: Users can earn points or tokens for referring new users to the platform. These points or tokens can be redeemed for rewards, such as access to premium content or merchandise.

Competitions: Users can compete against each other in various competitions, such as writing contests or photography contests. Winners can receive rewards, such as tokens, merchandise, or access to premium content.

Micro-tipping: Users can send small amounts of tokens or cryptocurrency to other users as a way of showing appreciation or support for their content. This can help to incentivize users to create high-quality content and engage with the platform.

Partnerships: Leo Finance can partner with other businesses or platforms to offer exclusive rewards or promotions to its users. This can help to attract new users to the platform and increase engagement among existing users.

When you find something that you like, you have to dig deeper. As I said, GPT and any other AGI fill in many blank spaces you leave out, so in order to get a better query, you need to be more specific and encourage the robot to be more thorough.

Eric : Number 5, "partnerships" caught my attention. How can Threads be used as a collaboration tool with other crypto projects, but also normal web2 companies?

And so on...

But you can even ask the AI to look for loopholes in the concepts you have, and look for ways to rig the system.

Eric: Great insight. As you know, polls use custom_JSON to track who votes, what and when, but this makes Polls on Threads on Leo Finance an easy to rig process, because anyone with a hive blockchain wallet can vote on any poll. What if you want to pick the best art piece by making a poll, but that person tells all their friends to vote for them? Doesn't this make Polls a popularity contest instead of an unbiased, tamper-proof process? How would you avoid a poll being rigged?

You can even play around with code at a very basic level


But you definitely have to correct it, maybe even call it a dummy bot.


Now I'm not telling you to treat the AI as a human, but the more specific and attention to detail you have in your prompt, the better the bot will understand what you want. After all it is trained by humans, so having human interactions with it can take you a long way.

Eric: This is great, but I'm worried that users would get discouraged from seeing polls they cannot vote on. How would this issue be fixed? The users can't vote on the poll, so would it be advisable to not show them the poll, or would it be better to show a message telling them the reasons? Or better yet, give me some options about this

And all you need to do is start with a basis, then go into general topics, extract good intel, correct where you need to, and then go into the specifics:

Eric: Now considering how Threads work, and taking into account how the hive ecosystem work, tell me more ways to use Polls on Threads that overall benefit the hive blockchain, the leo finance ecosystem, and the Threads feature.

... ... ...

An Boom, it's not a matter of if, but of HOW people integrate AI as a personal assistant to help them do their jobs better, faster and in an easier way.

AI won't take your job, but those who learn how to leverage AI definitely will.