How we feel about using ChatGPT on AIPs

With the surge of AI, many AIPs are in similar chat bot formats where a computer is writing the texts. Although easier to finish an AIP, I think writing it yourself gives you more awareness of what you are proposing, especially when asking for a fund. Of course, there are great authors with written skills, so wanted to know your thoughts.

Do you support authors using AI?

  • Yes
  • No
0 voters
1 Like

I don’t see a problem with it. I mean, we use various tools to do specific digital things; so why not use the power of one more tool? Yes - as a technologist, I recognize the limits of using any tech, as well as the pros and cons; yet still those don’t preclude the advocacy for simplicity and progress.

People can use it, but just add a disclaimer that they used AI in the writing.

Another great use of AI is to read HEX data before signing contracts. I can read HEX data but many people can’t. It also helps you learn how to interpret HEX data.

I used perplexity ai tool to do this.

1 Like

Explain why that is.

If I go searching for something on the Internet, I don’t have to add a disclaimer that I found it by searching, what search engine I used etc.

Let’s take it one step further. What you’re implying is that people should know also disclose what tools they use for other digital work e.g. art, audio etc. Why is AI any different? If I use Photoshop to make an image, and I use its filters to make art look a certain way, do I then need to disclose that? It’s what you’re suggesting by saying there should be a disclosure that AI tools were used.

Yes - but you can’t just “learn how to interpret HEX data” because in order to do that, you must first know what the underlying HEX data even means. And the use of AI simply does the interpretation. As you’re aware, it’s basically like translating one language to another. e.g. here is an example of interpreted HEX data and this is the entire lexicon. Unless you’re already familiar with HEX, whatever the AI interprets it to be, is patently useless.

Here is an example. It isn’t perfect but certainly helps one interpret and learn more about HEX transaction requests.

translate this into plain english

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

This appears to be a hexadecimal representation of blockchain transaction data, likely for an Ethereum or similar blockchain. Translating it to “plain English” involves interpreting the different parts of the transaction. Here’s a breakdown of what this data likely represents:

  1. Function Signature: 0x6a761202
    This is likely a function identifier for a smart contract.

  2. Contract Address: 0x4d224452801aced8b2f0aebe155379bb5d594381
    This is the address of the smart contract being interacted with.

  3. Transaction Data: The long string of zeros and other hexadecimal values represent various parameters and data being passed in the transaction.

  4. Method ID: 0xa9059cbb
    This is likely the method ID for a token transfer function (specifically, the “transfer” function in many ERC20 tokens).

  5. Recipient Address: 0x2136d9bb81a20fc8923f7e84d0653d83d983e02b
    This appears to be the address receiving the transfer.

  6. Amount: 0x8ac7230489e80000
    This hexadecimal value likely represents the amount being transferred (10 tokens if this is using 18 decimal places, which is common for many tokens).

  7. Signature: The final long string of hexadecimal characters is likely the digital signature authorizing this transaction.

In plain English, this transaction data could be interpreted as:

“A request to transfer 10 tokens from a specific account to the address 0x2136d9bb81a20fc8923f7e84d0653d83d983e02b, interacting with the smart contract at address 0x4d224452801aced8b2f0aebe155379bb5d594381. The transaction is signed and ready to be processed on the blockchain.”

Please note that without more context about the specific blockchain and contract, some details might vary, but this gives a general interpretation of what this data represents.

Right. Now see below.

That was precisely my point: “Unless you’re already familiar with HEX, whatever the AI interprets it to be, is patently useless.”

Regarding the disclaimer, in academia if you use AI, you need to cite or acknowledge your use of AI. It is different to using a search engine and then writing in your own words. It is like citing a source, like a book, news article etc. Here are links to policies from various universities.

How do I cite AI correctly? - Artificial Intelligence (AI) and Information Literacy - Research Guides at University of Maryland Libraries.

https://libguides.brown.edu/c.php?g=1338928&p=9868287

Also some interesting medium news.

I am aware of that context. But the discussion here in the community is about requiring authors to disclose their use of AI in composing their proposals. That’s a completely different standard than that of academia, media publications etc.

Plus, AI isn’t prohibited. It’s a case by case use case. e.g.

https://www.sciencedirect.com/science/article/pii/S2666990024000120#:~:text=AI-powered%20writing,[7].

Some examples:

AI in academia: An overview of selected tools and their areas of application
Guidelines for Using AI Tools in Writing and Research - Walden
ARTIFICIAL INTELLIGENCE (AI) IN ACADEMIA: RESOURCES FOR FACULTY - Texas U

More importantly, in case you weren’t already aware, for publications outside of academia it’s more about plagiarism than about original content seeing as LLMs are trained from human input.