After being excited about the launch of GPT-4 in early March 2023, now the crypto community is reportedly getting excited by the prospect of automating their trades through future technology built with the GPT-4 API.
What do you think it’s like? Check out more below!
Get to know BabyAGI and AutoGPT
Recently a spate of apps built on OpenAI’s GPT-4 API made the crypto community excited with a design that led to the development of a fully autonomous and self-correcting cryptocurrency trading bot. The two newest apps, named “BabyAGI” and “AutoGPT”, have been receiving particular attention, with many users trying to build crypto trading apps on top of them.
Furthermore, the big idea behind both apps involves task management for GPT-4. Currently, GPT-4 is believed to excel in natural language processing, due to the proven usability of the ChatGPT interface. Even so, GPT-4 cannot remember specific information.
Based on the conversation that took place in a tweet, the ValleyUnder account shares its experience in using AutoGPT to combine the best indicators for crypto trading.
Disadvantages of GPT APIs
According to a Cointelegraph report, these applications built on top of the GPT API are essentially limited to using a single session, which means the model cannot remember information from previous interactions. This has to do with the amount of data, referred to as the number of “tokens”, required by each query, as well as the GPT’s tendency to hallucinate, a problem that can occur when the number of tokens increases.
Basically, users can start with an empty record or without any data every time they perform a query on the machine. When it comes to building self-correcting crypto trading applications and historiographical analysis that adapts to real-time market conditions, the most powerful bots built on top of the GPT API still require intensive human supervision.
This means that even though GPT technology can assist in processing information and analyzing market trends, these bots still require human supervision to avoid errors or failures in automatic crypto trading.
While it still has some drawbacks, some savvy developers may have discovered a potential method to circumvent these limitations by building applications that take advantage of GPT’s ability to generate code and connect to external sources.
Of the many trading bots, the goal of certain specialized applications looks to go beyond automating crypto news aggregation or teaching machine learning agents how to recognize downturns. AutoGPT, for example, uses GPT-4 to generate code and then exploits GPT-3.5 to create what appear to be virtual artificial memory spaces, where information is combined and shuttled between them.
Another effort, BabyAGI, combines GPT-4 with LangChain, a coding framework, and Pinecone, a vector database, to coop new agents to complete complex tasks without losing focus on the original goal. Further, both applications could have the potential of serving as backbones for multi-agent AI applications capable of top-down management of crypto portfolios based on just a simple language request.
While there doesn’t appear to be an app specifically designed with the cryptocurrency market in mind, the team from Cointelegraph have seen several attempts on social media and GitHub to adapt one or both of the apps for autonomous trading.
The development of AI technology cannot be avoided, this also applies to DWF Labs, a mockup maker and investment company, which continues to inject funds into several AI cryptos, including CryptoGPT some time ago.
