Regulation, AR and RAG

Regulation and Markets

Markets respond well to the resolution of uncertainty, especially in the face of a potentially delayed or contested result. Consistent with previous Republican outcomes, US equity markets responded positively to the Trump Vance outcome. The deregulation wave is likely the most potent positive factor for the majority of the stock market over the next year. The business sector is desperate for a deregulation regime that can re-open the door for M&A and cut compliance costs across the board. You can see this reflected in market prices for investment banks, which were all up double digits. Regulation cuts are like tax cuts; they reduce expenses and remove substantial open-ended financial and legal liabilities based on ever-changing policies. Regulation cuts will also impact the Inflation Reduction Act (IRA) and its subsidies for the green energy / clean energy movement.

The polling industry dramatically underperformed betting markets (Polymarket or Kalshi). $3.6 billion in bets were placed on the presidential election outcome on Polymarket, which does not operate in the US, but US citizens can still access it via crypto and VPNs. The largest winners each walked away with $20m in profits. The betting markets were the first to “call” the result as the market shifted to 90%+ certainty late Tuesday night and well before official announcements. Predictive betting markets are here to stay. We will need to keep a close eye looking for signs of insider trading and manipulation. The crypto community’s overall belief is that blockchain transparency allows for crowdsourced monitoring of suspicious activity. There is evidence that this watchful collective eye is already functioning correctly in betting markets. I’m a little suspicious there will be rampant manipulation and insider trading, but we will have to wait and see.

Bitcoin hit an all-time high during the election evening. Trump has promised to remove SEC chair Gary Gensler on his first day in office. Gensler has been hostile to the crypto industry overall. Overnight, the headwinds for crypto have turned into fairly significant tailwinds. Trump also agreed to pardon Ross Ulbricht, the original Silk Road founder, who has been serving a life sentence.

Traditional scripted media also significantly underperformed unscripted podcasts. All politicians must embrace long-form, unscripted interviews to be optimally competitive in future elections. Independent podcasts are a significant media format for news, information, perspectives and entertainment.

Innovation Updates

A few things caught my eye over the last week.

Meta’ AR glasses

Marques Brownlee reviewed META’s augmented reality glasses. The next version of the metaverse is still a ways off. These glasses are a clear step in the right direction, but in their current alpha configuration, they would need to retail for $25,000. There is a long road ahead to shrink all components and reduce costs by at least 10x. They will get there, but it may be a while. Still, it’s a very impressive technology.

RAG AI – Retrieval-Augmented Generation AI

I’ve recently learned how powerful the RAG framework is for the future of AI applications, so I thought I’d mention it here. You may see it more often as a buzzword going forward.

The biggest issue with large language models (LLMs) is hallucination. To address this, RAG combines LLMs with external databases, enabling the AI to retrieve critical data from trusted sources rather than generating it from scratch. The AI uses that reference database to find and incorporate relevant information directly into its responses. The result is a significant reduction – often near elimination – of hallucinations, allowing the AI to excel conversationally without sacrificing accuracy on critical details.

This is extremely useful in fields like customer service, where product names, customer names and addresses need to stay consistent across conversations. Precision is important for any discussion about medical lab test results, financial reports, legal case studies, historical events and so on.

The next time you discuss AI hallucinations, ask about RAG to see if it has contributed to an overall improvement. You can read about RAG on NVDA’s blog if you want a deeper dive.