This article is based on a talk by Jason Mars at the Michigan Technology Leaders Summit, presented by SIM Detroit, on the topic of Artificial Intelligence: Actual Use Cases.
What criteria do you use to prioritize AI projects in your portfolio and how often do you ensure alignment with broader business objectives?
In the last year, we’ve partnered with companies to bring two generative AI solutions to production, observing key drivers across multiple sectors. These drivers can be categorized into two main themes.
Firstly, companies see AI as an opportunity to scale productivity, market output, and business efficiency. On the other hand, they perceive AI as a risk, navigating a new competitive landscape where survival depends on capitalizing on AI opportunities. This is particularly evident in the financial sector, where companies recognize the need to compete in a more efficient market.
The launch of ChatGPT marked a significant shift in the market, with VC investment in AI skyrocketing from $2 billion to $14 billion within six months. Companies realize that to survive, they need to compete in a higher efficiency landscape. This realization has led to a surge in market and internal capability analyses to identify the safest starting points for AI implementation.
Financial & Business Use Cases of Generative AI
A notable trend is the democratization of AI, which has become a mandate rather than an option. Smaller, specialized models have emerged, offering cost-effective alternatives to large foundation models like GPT-4.
For example, we’ve worked with PocketNest, a Michigan-based company, to build a conversational AI focused on financial advice. By using smaller open-source models, we’ve significantly reduced costs while maintaining competitive quality.
Another innovative use case is TOBU, a product that allows users to attach memories to pictures through a conversational AI. This AI interacts with users about their experiences and the context of their photos, creating a personalized memory assistant.
What are some of the most significant challenges you’ve encountered in implementing these projects and what have you learned from it?
Cost remains a significant challenge when scaling AI solutions. While development costs might be manageable during the initial phase, launching a product to thousands of users can become prohibitively expensive. This has driven a deeper investigation into using challenger models like Mistral and Llama to balance cost and performance effectively.
Expertise within partner companies also plays a crucial role in the successful deployment of AI. Our consulting approach involves delivering IP and production-ready AI engines while navigating challenges such as bounding AI use cases to prevent liability and ensuring that AI solutions remain within desired parameters.
How do you assess state and federal regulations, security and the privacy?
As we consume more information through screens, the realm of AI expands. It’s essential to control this growth thoughtfully, ensuring AI remains a tool for good. The current phase of AI development involves creating tools to harness the full potential of these models, akin to refining raw ore.
Tools like Lang Chain, funded by Sequoia, exemplify the innovation in this space. These tools help solve bigger problems by enabling seamless interaction with large language models. Personalization remains a key focus, making AI more accessible and tailored to individual needs.
In conclusion, the landscape of AI is evolving rapidly, with companies seeking to balance opportunities and risks. Through careful analysis, cost-effective solutions, and innovative use cases, businesses can harness the power of generative AI to drive efficiency and personalization in unprecedented ways.
