The rapid evolution of Large Language Models has led many to believe that artificial general intelligence is imminent. We see systems drafting correspondence, writing code, and even passing complex exams. However, a significant development in the technology sector suggests that the path to true intelligence is far more complex than simply scaling existing models.
Yann LeCun, the former chief artificial intelligence scientist at Meta, recently announced that his new startup, Advanced Machine Intelligence, has raised over one billion dollars to develop what he describes as ‘world models’. His argument is clear: the idea that current language-based models will reach human-level intelligence is, in his words, ‘complete nonsense’. This shift in focus highlights a persistent set of limitations that businesses should consider when evaluating their technical strategies.
The Problem With Language-Based Intelligence
The primary issue with current artificial intelligence is that its reasoning is grounded in language patterns rather than a genuine understanding of the physical world. While these models are revolutionary tools for productivity, they often struggle with the deep, contextual problem-solving that defines a human expert.
One of the most significant barriers to adoption in critical sectors is ‘hallucination’—the tendency for models to confidently state information that is entirely fabricated. Emerging research suggests that while developers are identifying the issues causing these errors, they remain a fundamental hurdle. Even with advancements in memory management, current models often lack the persistent memory and reasoning capabilities required for high-stakes decision-making.
The Average Output Gap
Real-world benchmarks continue to show that artificial intelligence is often below average when compared to a human specialist. In complex engineering and technical fields, tests demonstrate that while machines can assist with rote tasks, they frequently fail when faced with multi-step, nuanced problems.
This creates an average output problem. Because the technology relies on historical data and patterns, it tends to produce results that are safe and conventional but lack the creative intuition required to solve bespoke or novel challenges. In a world where cyber security threats are constantly evolving, relying on average protection is often insufficient.
Assistance Versus Replacement
To understand the future of this technology, it is helpful to look at the history of the workplace. Computers did not eliminate the need for administration; they replaced the manual typewriter and the physical filing cabinet. They transformed the role of the traditional assistant into a digital-first position, increasing efficiency without removing the need for human oversight.
Similarly, artificial intelligence is currently replacing junior assistant tasks. It can suggest code snippets, summarise long reports, or draft basic outlines. This serves as a powerful accelerant for senior professionals who already possess the expertise to verify and refine the output. It is a tool for assistance, not a total replacement for human judgement.
Why the Expert Still Matters in Cyber Security
In the realm of cyber security, the stakes are far too high for ‘good enough’ solutions. An artificial intelligence may be able to suggest a general security policy, but it cannot comprehend the unique cultural and operational nuances of your specific organisation. It lacks a sense of responsibility and a holistic understanding of risk.
A human expert brings years of experience, ethical judgement, and the ability to anticipate creative attacks that do not follow established patterns. This is why leading scientists, including the former chief at Meta, are now looking beyond language to build systems that can reason and plan more like a human being.
How Vertex Can Help
At Vertex, we embrace technology that enhances our ability to protect your organisation. We use advanced tools to ensure our services are thorough and efficient, but we do not allow a machine to have the final word. Our expert penetration testers and cyber security specialists provide the critical human oversight required for keeping your business truly secure.
If you are looking to integrate new technologies into your business safely, or if you require a high-quality audit of your current security posture, consider reaching out to the team at Vertex. We can help you navigate the complexities of modern intelligence while ensuring your defences remain robust and effective.