Three ex-DeepMinders are working on creating A.I. to select stocks and cryptocurrencies.

• In January, Martin Schmid, Rudolf Kadlec, and Matej Moravcik departed from DeepMind to establish EquiLibre Technologies in Prague, the Czech Republic.
• IBM's trio developed an AI called DeepStack, which became the first AI capable of defeating professional poker players at heads-up no-limit Texas hold'em poker.
• Instead of playing poker, our algorithms will engage in algorithmic trading, as suggested by Schmid to CNBC.

Three ex-DeepMind staff members are attempting to develop a machine that can detect and invest in rising company stocks and cryptocurrencies.

In January, Martin Schmid, Rudolf Kadlec, and Matej Moravcik left DeepMind to establish EquiLibre Technologies in Prague, relocating from Edmonton in Canada.

In 2017, the trio, who had previously worked at IBM, created DeepStack, an AI that surpassed professional poker players at heads-up no-limit Texas hold'em poker.

Financial markets are now being considered for the application of these concepts.

Instead of playing poker, our algorithms will engage in algorithmic trading, and we are exploring the crypto market, as Schmid shared with CNBC.

The objective is to employ reinforcement learning to teach an AI system how to purchase and sell shares for profit. Reinforcement learning entails rewarding an AI system for accomplishing a specific goal (such as winning a chess game or detecting a tumor on a mammogram).

Schmid stated that he is not worried about regulators restricting the technology, as other companies are already doing similar things. In fact, EquiLibre Technologies will be competing with companies like AI algorithmic stock picking products Candlestick and Yuyostox.

Schmid stated that while most trading is currently algorithmic, the goal is to improve upon existing algorithms.

EquiLibre Technologies aims to either utilize the AI it creates to establish a new hedge fund or sell it to a major institutional bank or another investor in the future.

The advisory board of EquiLibre Technologies comprises two prominent DeepMind staff members who are renowned in the field of AI.

Michael Bowling, head of DeepMind's Edmonton office, and Richard Sutton, co-author of DeepMind's "Reward is enough" paper, claim that rewarding an algorithm for performing tasks it is programmed to do will eventually lead to it exhibiting signs of general intelligence.

EquiLibre Technologies has secured the largest-ever seed round in the Czech Republic, with venture capitalists providing backing, according to Schmid, although the exact amount remains undisclosed.

"Start-ups are plentiful, but VCs struggle to identify the promising ones," he stated.

A growing number of ex-DeepMind entrepreneurs, including Schmid and his co-founders, are securing funding from venture capitalists.

Schmid stated that if one worked at Google, DeepMind, and other esteemed companies, they likely possess competence.

Staff at DeepMind are likely to have a network of tech contacts that could be recruited.