Source: Cointelegraph

Translation: BitpushNews

Imagine if ordinary people could write smart contracts in their native language, it could bring "billions of new blockchain users."

Emin Gün Sirer, founder of Ava Labs and creator of the Avalanche blockchain, said: "Smart contract programming is difficult. Capturing 'intent' is hard, writing code is tricky, and verification remains a challenge."

Sirer stated at the recent Cornell Blockchain Conference in New York City: "This is the main issue hindering the flourishing of smart contracts."

But imagine if artificial intelligence (AI) like ChatGPT or Llama could help you with all the smart contract coding work, lawyers might soon become the main authors of smart contracts rather than programmers.

Even more interestingly, ordinary people will be able to write smart contracts in their native language in a peer-to-peer manner, making it as simple as writing a bank check.

At the Cornell Blockchain Conference held in New York City on April 26, Sirer further elaborated on this future vision:

"You are all familiar with the process of writing a check, right? That is the most complex financial instrument our banks provide. Currently, our choices are quite limited. For example, I can write a check to 'John' stating 'I hereby transfer $5000 to John'—that's basically it."

But if we could perform the same operation and add additional conditions, Sirer continued with an example:

"I hereby offer $5000 to John, with the condition that he must raise $5 million by the end of September to shoot a movie. If he fails, I want my money back.' I can state it directly like this."

Sirer mentioned that this scenario he envisions may take another 5 to 10 years to materialize, but if it does, it will be "transformative."

"We will be able to onboard billions of new (blockchain) users."

Currently, most smart contracts are written using the programming language Solidity, which is not widely known even among the programmer community.

However, Avalanche has already begun developing a new virtual machine in the field of artificial intelligence and blockchain technology integration. Sirer stated that this virtual machine "supports programming using natural language. You can choose to use English, German, French, Tagalog, Chinese, or any natural language your mother taught you to write your programs."

Challenges Remain

But is this just a pipe dream? Before what Sirer calls "coin-operated agents" can be widely adopted, many issues need to be addressed.

Firstly, legal aspects must be handled properly. Keywords and terms need to be precisely defined. For example, before using terms like "software" in smart contracts, a very detailed definition may be required. Completing this work may take some time.

Additionally, whether artificial intelligence "agents" are ready to draft legally binding documents is still a question. After all, large language models (LLMs) sometimes produce erroneous information, a phenomenon sometimes referred to as "hallucinations."

Sirer pointed out: "Everyone has probably seen [images of people with] seven fingers generated by artificial intelligence."

However, some of these issues can be addressed in the short term. For example, transactions can be simultaneously submitted to ChatGPT 3.5 and Llama, requiring both agents to reach a consensus before smart contract solutions are adopted. "This way, at least you have two trusted executors, and you are taking the intersection of their actions."

"Instructions for writing smart contracts in simple English"

Some have also stated that smart contracts capable of handling "natural" language have already emerged—or are within reach.

Sam Friedman, Chief Solutions Architect at Chainlink Labs, told Cointelegraph: "In today's world, there already exists the ability for everyone to become their own programmer."

Friedman mentioned a project in the Chainlink BUILD program, the Council artificial intelligence agent framework developed by ChainML, which "allows developers to provide simple English instructions and immediately convert them into executable code. Currently, these codes are mainly Python or SQL, but a single upgrade is all that is needed to achieve smart contract functionality."

He also mentioned the "Ask AI" feature in the Chainlink Developer Center, where when users submit requests in simple English, the feature can return a complete smart contract. For example, a user can request: "Show me an example of a smart contract that allows users to request a random number between 1 and 10, using Chainlink VRF (Verifiable Random Function)."

However, current products may not yet fully realize all the functionalities Sirer envisions. Friedman stated in an interview with Cointelegraph that there is still much work to be done in the future.

He said: "Although artificial intelligence agents for writing smart contracts have emerged, they have not reached their final form. They still need further improvement in quality and complexity. To achieve this, artificial intelligence models need to be trained with continuously updated content."

Nevertheless, Friedman can still envision some scenarios where multiple artificial intelligence "agents" playing different roles collaborate to manage the lifecycle of smart contracts. For example:

One agent helps define requirements;

Another agent is responsible for writing smart contracts;

Yet another agent performs quality and security testing;

Another agent is responsible for deploying contracts;

The final agent continuously maintains or adjusts parameters or global variables in the contract after deployment.

At the Cornell Blockchain Conference, Sirer acknowledged that there are still "huge unknowns" and progress requires guidance from all aspects, involving those who have "both technical and ethical and humanistic backgrounds."

Room for Lawyers?

During the Q&A session after the speech, someone asked Sirer, if Solidity language is no longer needed and ASCII text can be used for coding, who will write smart contracts in the future?

"So who would be the ideal programmers? At first glance, software programmers seem to have the upper hand. After all, they are already adept at using precise language to define and specify conditions.

"But those who truly understand how to use formal language to draft contracts are lawyers. Therefore, they will become the leaders in that field[...] those who can truly thrive in this system."

As for ordinary users, what about their situation? "I believe they are fully capable of defining basic transactions," Sirer said. In fact, he expects these new features to be built on a "democratic, open platform" that "anyone can use equally."

Overall, this new approach to smart contracts is very appealing and full of hope, "we cannot just set it aside," Sirer concluded.