What Does an AI Chatbot ‘Think’ About DeFi? We Asked ChatGPT

Over the past month, my Twitter feed has been inundated with all things FTX and Sam Bankman-Fried. Between the apologies and interviews (why is Sam still talking?), my feed has been all fraud and no fun.

That is, until ChatGPT came along.

This article originally appeared in Valid Points, CoinDesk’s weekly newsletter breaking down Ethereum’s evolution and its impact on crypto markets. Subscribe to get it in your inbox every Wednesday.

If you haven’t played around with the new chatbot from OpenAI you really should. From movie pitches to programming queries to poetry, the AI reliably spits out human-like, logical responses to nearly any prompt you throw at it. Speaking with it has me feeling like I did when I first swiped to unlock the screen of an iPhone.

It’s a funny thing, AI. Like crypto, artificial intelligence has been held up by some as a paradigm shift in how the world ticks. But unlike crypto, which has yet to yield many real-world applications beyond finance, it’s hard to push back against the AI-everything thesis. At some point – probably soon, if not already – it will be difficult to think of an industry that hasn’t been completely upended by machines that can think.

For this week’s Valid Points, I’ve asked ChatGPT to speak a little bit about the impact that AI will have on DeFi.

Am I lazy for doing this? Yes. Do I still think it's interesting? Also yes. Someone at CoinDesk had to figure out whether AI is going to take our jobs, and I’m happy to have been that someone.

The AI prompt

We offered the following prompt to ChatGPT: "Write a CoinDesk article about how AI might be used in decentralized finance." The bot returned six paragraphs. We've included the first five paragraphs verbatim.

We replaced the AI's concluding paragraph with its response to our follow-up prompt: "include risks as well as benefits." This yielded three paragraphs, which we again recorded verbatim.

The response by ChatGPT

As decentralized finance continues to grow in popularity, many are looking to artificial intelligence (AI) as a potential solution to some of the challenges facing this emerging technology.

Decentralized finance, or DeFi, refers to a system of financial transactions that are performed on a blockchain network. This allows for the creation of digital assets and the creation of smart contracts, which enable the execution of financial transactions without the need for intermediaries such as banks or other financial institutions.

One potential use case for AI in DeFi is the creation of more sophisticated and intelligent trading algorithms. These algorithms could be used to analyze market trends and make more accurate predictions about the direction of asset prices, helping traders to make more informed investment decisions.

Another potential application of AI in DeFi is in the creation of more advanced and efficient lending and borrowing platforms. By using AI algorithms, these platforms could automatically assess the creditworthiness of borrowers and set appropriate interest rates, reducing the risk of defaults and making the lending process more efficient.

Additionally, AI could be used in DeFi to improve the security of smart contracts and other blockchain-based financial transactions. By using machine learning algorithms, smart contracts could be automatically monitored and audited for potential vulnerabilities, helping to prevent hacks and other security breaches.

However, there are also potential risks associated with the use of AI in DeFi. One concern is that the use of AI algorithms in trading and lending could lead to the creation of "black box" systems that are difficult to understand and regulate. This could make it difficult for regulators to monitor and control these systems, and could also create potential risks for investors and borrowers.

Another potential risk is that AI algorithms could be subject to bias or manipulation, leading to unfair or inaccurate decision-making. For example, if AI algorithms are trained on biased or incomplete data, they could make decisions that are unfair or discriminatory. This could be particularly problematic in the context of lending and credit scoring, where the use of biased algorithms could lead to discrimination against certain groups of borrowers.

Overall, while the use of AI in DeFi has the potential to bring significant benefits, it is important for developers and regulators to carefully consider and address these potential risks. By taking a cautious and responsible approach, it may be possible to harness the power of AI to improve the capabilities of decentralized finance without creating unintended consequences.

Human here again

Pretty good, huh?

It took some trial and error at the beginning, but once I crafted the magic prompt, I think ChatGPT managed to string together a relatively convincing (albeit dry) argument for how AI might be used in DeFi.

That being said, I'm not sure ChatGPT is ready to take over this newsletter just yet. The bot sounds human, but it’ll frequently manufacture facts out of thin air, and I had to weed through some pretty weird responses before landing on something fit to print (ask me about the time it wrote – incorrectly, for reasons that probably don’t need explaining – that “Alameda Research is a quantitative trading firm that uses AI to make fast and accurate trades on decentralized exchanges”).

If you are interested in reading a human take on how AI intersects with DeFi, CoinDesk has you covered. While ChatGPT covered a lot of ground in its article, it did miss some key applications, such as insurance, and key risks, including how on-chain AI could be used to manipulate markets or harm users through malicious MEV strategies.

My job is still safe, but one imagines that it won’t be long before ChatGPT is explaining – I hope with a bit more color next time – not only how AI could transform DeFi, but how it already has.