How Combining LLMs and LAMs Can Transform the Future of Business
We are in an era of rapid and ongoing digital transformation. Artificial intelligence has fundamentally and permanently changed the business landscape—and it will only continue in this direction. There is no going back to the way things were before, and I believe that we should embrace these changes with excitement and optimism, not fear and negativity.
AI tools can eliminate or significantly reduce the time you spend on boring, repetitive tasks, giving you and your team more free hours to fill with creative, interesting and fulfilling projects. Large language models (LLMs), including those powering ChatGPT and Bard, are now improving efficiency in areas such as content creation, data summarization and customer care. Large action models (LAMs) are not yet as widely available as LLMs, but they promise to be another game-changing technology. By learning to use LLMs and LAMs together, you can take advantage of their individual strengths and maximize their combined impact.
Understanding LLMs And LAMs
LLMs and LAMs are two distinct types of AI models:
• LLMs excel at comprehending and synthesizing data and generating human-like responses based on the inputs and instructions they receive. If you ask an LLM a question, it can provide a detailed summary of the massive information stores it has access to—though it can also “hallucinate,” or generate a false or inaccurate answer. LLMs are trained on huge datasets to understand a wide range of topics and contexts.
• LAMs are trained to understand human intent and perform actions, whether proactively or based on previously outlined instructions. An LAM doesn’t simply respond to queries; it can learn to mimic human tasks and autonomously execute basic functions.
Unlocking Future Opportunities With LLMs And LAMs
If an LLM serves as the brain, processing vast amounts of data and extracting meaningful insights, then an LAM acts as the hands, translating information into tangible actions and processes. When they join forces, they can take over mundane tasks traditionally performed manually by humans.
For example, you could train an LLM and an LAM to work together as a personal scheduling assistant. If you need to schedule a call with several people, the LLM could search your calendar for availability for an afternoon meeting in the next two weeks, then the LAM could send participants an email with proposed times, a booking link and a meeting agenda. Once the LAM is trained on this particular function, it would be able to replicate it on its own for future meetings based on similar parameters.
The convergence of LLMs and LAMs marks a new era of AI and automation. Organizations in all industries could use these tools in tandem to enhance productivity, decision-making and innovation. Here are a few examples of possible future applications:
• Marketing: Train an LLM to analyze and monitor Google ads to determine which campaigns are most effective for customer acquisition. Translate these insights into actions for an LAM to execute, such as refining search terms or increasing the budget for certain campaign criteria.
• Supply chain: Use an LLM to track product inventory, taking into account factors such as market demands, seasonality and marketing campaigns. Train an LAM to automatically order items in advance to avoid running out of stock and ending up with back orders.
• Social services: Utilize an LLM to navigate government websites and summarize complex information about benefits and grants for low-income or underserved populations. Design an LAM to filter for eligibility requirements and apply for aid, job openings or other opportunities.
Navigating The Path Forward
LLMs and LAMs are still in their early stages of commercial development, but they are evolving rapidly, and we can expect big advances in these technologies in the near future. Begin exploring AI applications now, and prepare your organization for the next chapter.
Start Small
Identify low-hanging fruit—manual business processes that could be easily optimized with automation for a meaningful impact. Implement AI tools to resolve the smallest problem you can find. Celebrate your first win and then look for a bigger challenge to tackle. Keep testing and experimenting, gradually scaling up to more complex problems with bigger payoffs.
Build Knowledge And Expertise
Invest in cultivating a deep understanding of LLMs and LAMs, their capabilities and their limitations. Learn about best practices for protecting the security and privacy of your data in LLMs. Research options for building or training LAMs in cloud computing environments you are already using. Model a culture of continuous learning and experimentation with AI.
New technologies have always helped humans evolve and thrive. In the future, I don’t think it will be humans versus AI. It will be people who use AI to make their lives and businesses better versus people who don’t. The people who choose not to embrace AI, as with any other technological innovation, will struggle to keep up. We will inevitably face challenges moving forward. Some jobs will become obsolete as new opportunities emerge; some tools will flail while others flourish. Proceed with caution and thoughtfulness, but stay open, curious and optimistic about what will come next.
