Understanding how AI is impacting the world and will continue to do so is important for everyone, but particularly for people with disabilities and those that care for them.
A JP Morgan report evaluating the impact of Trumps tariffs noted that “Investors have been laser-focused on tariffs, but tariffs aren’t the whole market story.”. That AI LLMs “are being used far more aggressively” which is compounding over time and increasing productivity.
JP Morgan wrote:
“We believe a revolution is underway and we’re only in year three. In the long run, AI will drive higher productivity, leading to higher gross domestic product (GDP) growth. Tariffs raise the hurdle; AI raises the bar. Overall, higher productivity can lead to incremental revenue and increased government revenues.”
My dad brought to my attention a graph image of the S&P 500 for the past year. The graph shows that in February 2025 when Trump implemented tariffs the market dropped from 6,117, and it kept dropping to a low of 4,982 on April 8th. It wasn’t until April 21st that the S&P 500 began to climb again, and it’s been climbing ever since, at 6,391 on July 29th 2025.
When I reviewed the S&P 500 data I realized it correlates to major developments and releases in AI. Google released Gemini 2.5 Pro Experimental on March 25th, which was a home run, they crushed LLM benchmarks and leaderboard rankings. OpenAI released their 03 model on April 16th, nearly matching Gemini 2.5 Pro Exp (the 2.5 Pro release was really that good). Within the time frame of 1 month, AI went from useful to very useful; it genuinely became better than most humans at most things. In subsequent months Google and OpenAI improved their models further, with Gemini 2.5 Pro Preview on June 5th and a stable “generally available” release June 17th. OpenAI released 03 Pro on June 13th.
Claude 4 released in May. As did Abacus.ai Deep Agent and other autonomous AI Agents that had more capabilities than Gemini, GPT, or Claude’s base models. These agents could solve more complex problems on their own, help build software, perform legal and medical analysis, etc. FutureHouse released their scientific research AI in May as well, which was made openly available and free to use, and has proved very capable and useful in medical and scientific endeavours.
The N8N platform helps automate tasks and create AI Agents. As AI models improve the capabilities and usefulness of the N8N platform improves as well. The N8N platform lets people without a coding background perform coding tasks, build complex AI Agents, and create systems and complete projects that they otherwise couldn’t. They can do things that hadn’t been possible to do prior to AI and N8N. The productivity of non-coding humans rapidly increased with N8N and each major AI model release.
The Model Context Protocol developed by Anthropic was released open source and adopted by all major AI LLMs. The MCP lets one connect databases and other tools to AI LLMs efficiently, greatly expanding their capabilities. MCP servers really started to take off in the March-June timeframe, and are expected to expand much more aggressively as MCP is maturing and gaining acceptance. In example, the Free Law Project’s CourtListener case law database (a freely accessible data of all U.S. Case law) has an API that lets one search it. This API can be connected to AI LLMs, but it’s difficult and technically challenging (I can’t do it myself).
With MCP servers it would be made rather simple to connect AI LLMs to the CourtListener Database. New projects on GitHub were made in June trying to do just that. The official developers at the Free Law Project have been waiting for MCP to mature more before they began to develop an MCP server, but seeing that the community is moving ahead without them, FLP devs are considering making a beta MCP server with no promises on stability or function. I believe that many other organizations and nonprofits have shared in the thinking and approach of the FLP, and that over the next 6 months many databases of very useful information will be made available to AI LLMs through official MCP server support by groups that maintain and provide that data. Some of those databases will be paid, but many will be free.
Frankly, I’ve been wondering all year when Google will let Gemini directly access its Google Scholar case law database. AI can be so much more than it is by simply connecting it with existing freely accessible databases.
I believe my analysis of these matters indicates that the release of AI models and other AI related developments vastly increased productivity and bolstered market growth, offsetting the negative impact of the tariffs. The AI revolution might fully explain the S&P 500’s improvement. I believe we’ll continue to see more market stability and growth over the next 3-6 months (maybe longer). But one should also expect turbulence, as there will be job loss, and multiple industries collapsing due to becoming obsoleted by AI. Many who lose their job will end up creating new companies and new jobs using AI, creating more market growth, while others remain unemployed. AI related job loss will be a challenging problem, but likely not a disaster.
I picked 3-6 months as it matches the observed historical trend from my analysis. OpenAI will release GPT 5 in August. Google is cooking Deep Think and Mariner, and presumably more Gemini releases to respond to and compete with GPT 5. Anthropic has been focused on a gradual path of expanding Claude capabilities – base Claude isn’t that impressive, but a customized Claude with lots of integrations and tools, people are getting some serious work done with that. Something that Gemini and GPT have both neglected.
While the focus of my analysis has been on how AI has rapidly developed and the narrow time frame of multiple releases, one should also consider how these more capable AI models not only bolster human productivity by helping us complete tasks, but they accelerate the rate at which humans can learn new skills. While some humans use AI so it thinks for them, others use it to help them learn and improve their thinking. AI outperforms most doctors, but it also helps those doctors improve their knowledge and capabilities at a faster rate. As for me, my legal work wouldn’t be possible without AI, and as AI develops my ability to Defend The Disabled will greatly increase. Which is important, as the ‘job’ of Defending the Disabled has been one that most people are refusing to do, even when it’s literally their job to do it.
Defend The Disabled 