Patthawikorn Boonin, a Thai man, recently made headlines after winning nearly $60 in a lottery with the help of an AI chatbot called ChatGPT. ChatGPT is a sophisticated AI language model that’s designed to converse on various topics, but predicting the future is explicitly not part of its training.

Despite this limitation, Boonin managed to find a workaround that enabled him to use ChatGPT to generate four two-digit numbers that he played in the Thai Government Lottery. To his surprise, the number 99 was drawn, earning him his cash prize.

The details of how Boonin achieved this feat are relatively straightforward. He got the AI to generate four, two-digit pairs: 57, 27, 29, and 99, which he then used in the lottery. However, this raises some interesting questions about the concept of randomness and the role of artificial intelligence in predicting lottery outcomes.

The core of any non-rigged gambling game is based primarily on randomness. If outcomes were predictable, no one would bet on losing options. However, there are nuances to the concept of randomness. True randomness involves unpredictable events that occur naturally in our universe, such as radioactive decay or the day we die. Algorithmic randomness, on the other hand, simulates random events using mathematical algorithms called pseudorandom number generators (PRNGs).

While PRNGs create number sequences that seem random, they are based on deterministic formulas and are, therefore, somewhat predictable. As PRNGs rely on initial values or seeds, artificial intelligence could, in theory, predict these algorithms’ outcomes by analyzing vast amounts of generated data. By detecting hidden patterns and potential correlations in number sequences, AI could refine its model to better predict the next pseudorandom algorithm-generated number.

However, predicting lottery winners is highly unlikely. Lottery numbers are drawn randomly, and the systems are designed to ensure that the process is as unpredictable as possible. To predict lottery winners, an AI would need to be able to predict true randomness, which is fundamentally impossible.

We asked GPT-4, a more advanced AI language model than ChatGPT, if it would ever be able to predict lottery winners. It responded by saying that the feat was highly unlikely. “To predict lottery winners, an AI would need to be able to predict true randomness, which is fundamentally impossible,” it said.

In conclusion, while Patthawikorn Boonin’s use of an AI chatbot to win a small sum of money in the lottery is interesting, it does not signify a breakthrough in the ability of AI to predict lottery outcomes. The limitations of predicting true randomness mean that lottery numbers remain unpredictable and that AI is unlikely to be able to provide a reliable way to win the lottery.