DeepSeek sent shockwaves through Silicon Valley with the release of its new AI models which it claims cost a fraction of the price of Western counterparts such as OpenAI’s ChatGPT.

As a result, US tech stocks tumbled as investors questioned the level of spending in the sector amid these new revelations.

However, it is claimed by some, including Donald Trump’s AI czar David Sacks, that DeepSeek was able to pull off the feat by a process of ‘distillation’ – which means training models using existing technology.

Reflecting on the news during the latest episode of iGaming Daily, Cristian Barbosa, Founder of InsightPlay.AI, said: “It’s like using something to build something completely new which is efficient because you don’t have to train it from the beginning.

“However, it’s a different thing to train a model and then run it at scale. What’s expensive is the inference part which means whatever happens after you as a user give it a query. All that [graphics processing units] power is what is very expensive to run. 

“It’s going to change a lot of things in terms of the economy of using this technology but not as drastically as everyone thought at the beginning, so clearly there was a market overreaction last week.”

In a busy month for AI, Donald Trump also revoked previous President Joe Biden’s AI Safety Bill during his first days in office, removing the guardrails surrounding AI development in the US.

Although this has positive implications for economic growth related to technology, Barbosa cautioned against the potential dangers of what he describes as “the most disruptive technology since the invention of the internet”.

“There are two ways you can use AI,” he explained. “The good way is improving people’s lives but the bad is using it for bad purposes which is mainly control of populations, decisions and democracies. It’s a very complicated time in terms of global politics and what I’m seeing is that the gloves are off, this is going to be a fight. This has the potential to change the world and they know that.”

Transformative potential in iGaming

Later on in the episode, Barbosa also considered the potential of AI to transform the user experience for iGaming players.

He said: “We’re going to see a lot more personalised experiences at scale. If your marketing manager is still talking about localisation, you’re late to the game. 

“The name of the game right now is hyper-personalisation. That goes for everything before you get to the operator or with the operator so that you have the most personalised experience [possible]. Everything is going to be custom and tailored to your preferences as a user.”

One area where AI has been leveraged by operators is to power customer service chatbots, however, these have been met with dissatisfaction from some players. 

On the horizon to replace the chatbots are conversational AI agents that can interact with players to help solve issues.

“These chatbots used to work on flows of logic and if you drifted away from that you break it,” said Barbosa.

“What we’re building is AI agents which are autonomous and they make decisions for themselves and have memories. So, they’re not only going to work for hyper-personalisation but actually solve customer issues at scale with the largest capabilities of customer service, terms and conditions, local regulations and compliance.

“It just works in a much more sophisticated and competent way than traditional solutions.”