Surviving the AI Storm: How Large Language Models are Reshaping Software and Shielding Hardware

The rise of large language models (LLMs) like GPT and tools such as Baby AGI and AutoGPT has brought a new wave of transformation to the technology landscape. In the midst of the AI gold rush, a curious paradox is unfolding. While software, the traditional realm of AI, finds itself under siege, the hardware and manufacturing sectors appear to be relatively immune, at least for now. However, as we venture further into the future, the implications could be far-reaching.

The advent of GPT-3.5, and more recently GPT-4, from OpenAI, along with other rival LLMs like Claude from Anthropic and Bard from Google, has taken the AI world by storm. The AI revolution is no longer confined to the abstract and esoteric domains of academia and tech companies. With the capacity to generate human-like text, from recipes to computer code, LLMs are becoming increasingly accessible to non-experts, entrepreneurs, and businesspeople.

The software sector, particularly SaaS (Software as a Service) companies, are feeling the heat. Traditional SaaS firms, with their entire codebase meticulously written by teams of software engineers, may soon find themselves competing with new entrants who've generated their entire codebase with GPT "Agents". GPT Agents are several instances of GPT that work together to achieve a certain goal:

Defense Mechanisms

While this might seem like a significant threat, established SaaS companies with certain advantages are not defenseless. Those that benefit from the 'network effect' (where more users lead to more value per user), have accumulated valuable data that increases the value of their service, or have built a powerful brand, are likely to be more protected. These advantages create a form of "lock-in" that can provide a buffer against the rapid advances in AI.

Businesses are scrambling to understand how to capitalize on this technology, with no obvious killer applications yet identified. However, what is clear is that these models have the potential to automate tasks once thought to be solely within the realm of human creativity and reasoning, from writing to creating graphics to summarizing and analyzing data.

This brings about a fundamental shift in the landscape of jobs and productivity. Whereas AI had previously automated routine, step-by-step tasks, the new generation of AI is poised to automate tasks viewed as creative, such as producing graphics. The economic implications are far from certain, with predictions ranging from an optimistic view of AI being a powerful tool that improves workers' capabilities to a pessimistic view of AI displacing well-paying jobs that require creative skills and logical reasoning (I'm in the latter group).

Hardware is harder to copy

In contrast, hardware companies are not as easily 'copied' by agents. This is because of the complexity and challenges of physical production, which are fundamentally different from the software realm. LLMs don't have arms and legs attached to them (yet!), and hardware companies require significant investment into capital assets (as opposed to SaaS companies investments going into paying software engineers).

Investment-wise, entities that are 100% SaaS and few of the defense mechanisms mentioned before (Shutterstock, Wix, Alteryx, Shopify), will see more threats from new AI created entrants than hardware companies.

However, it's worth noting that this is not a permanent state of affairs. AI is rapidly evolving, and it's only a matter of time before it makes significant inroads into the physical world, whether through advanced robotics or other forms of automation.

In conclusion, the ongoing AI revolution, fueled by the rise of LLMs, is bringing about fundamental changes in the software sector, while the hardware and manufacturing sectors remain relatively safe, at least for now. However, the future remains uncertain and the only constant in this landscape is change.