Will AI be superior to human investors?

ChatGPT, Sora, Bard and Co.: Artificial intelligence is still on the rise. And it could also change how we trade in the crypto market. Bots are already getting involved on a large scale. There are AIs that automatically trade for you. Or they help you make the right decisions. So far, none of this represents serious competition for human traders. Could that change in the future? BTC-ECHO spoke to Tomasz Gawron about this. He is the AI ​​head of XTB, a Polish exchange that has existed since 2002. Stocks, ETFs and cryptocurrencies, among other things, are traded on it. His conclusion: “Trading robots can come close to human levels.”

BTC-ECHO: How well do AIs currently trade cryptocurrencies?

Tomasz Gawron: This depends on many environmental variables and the data used to train the model. The performance of AI trading robots can approach human levels. Recent breakthroughs in so-called “zero-shot time series forecasting” make it easier than before. However, AI can be completely helpless when it comes to data resulting from new events. It should be noted that our team does not develop trading robots and we do not plan to suggest investment strategies to users. Rather, we want to provide tools that help users make informed and consistent decisions.

Are there special challenges for AI when it comes to cryptocurrencies – in contrast to stocks and the like?

Since cryptocurrencies are a relatively new type of asset, they appear to be heavily influenced by social media and new means of communication via blockchain. In my opinion, when analyzing cryptocurrencies, one must consider the technical fundamentals of crypto assets as well as other factors. Protocol changes, halving and other developments can be a source of high volatility that only affects a specific group of assets.

Considering that cryptocurrencies are new and rapidly evolving, it is a major challenge for AI systems to accurately predict what will happen during the high volatility events that commonly occur in the crypto world. The task of assessing market sentiment using social networks is also more difficult in the case of cryptocurrencies. There are more communication channels and communication runs faster than with other assets.

Do you think AI will eventually be able to trade better than humans? Will the human trader become extinct?

One could argue that algorithms are gaining the upper hand in certain specialized areas such as high-frequency trading. Well-designed AI-based systems can be more consistent and less biased than the often emotional human investor.

However, the patterns found by AI cannot always be generalized to the immediate future. In times of unpredictable events (black swan events), such as wars, disasters or sudden technological breakthroughs, the human investor has the opportunity to react better. In the immediate future, I see AI as just a tool for traders. However, that may change with another Chat GPT-like breakthrough in the future.

How will artificial intelligence change cryptocurrency investing in the near future? Give me one or two realistic examples.

We live in a society that relies on cloud computing. GPUs are the new currency. It is often said that smaller companies are “GPU poor” or “GPU hungry”. The development of AI crypto tokens that can be used to purchase GPU computing power and on-demand AI model inference services will therefore continue.

How can I as a crypto trader benefit from this new development?

It will be like having a team of analysts available to you as a trader at any time. One can create custom charts, indicators and trading strategies without any coding or deep technical knowledge. The trader can concentrate on making strategic decisions and reacting more quickly to the always volatile crypto market.

What problems will AI create for the trading sector?

I’m not sure if this is a problem, but AI could still improve the sector
make more competitive. Those who understand and master AI can gain an advantage over others.

One of the possible problems is massive market movements caused by AI models that misinterpret or hallucinate data. The more we rely on AI to analyze the endless stream of input data, the more vulnerable we become to its biases and errors.

Imagine that the AI ​​misinterprets some Twitter posts related to the recent price movement of a cryptocurrency. Then it makes a decision based on that premise and executes a sales transaction.

If this happens to a few AI systems with a high enough transaction volume, it could trigger a snowball effect as other automated systems could pick up on the resulting price movement.

Thank you for the interview.

This might also interest you

source site-52