Latest News & Resources


Blog Images

The Future of AI: Reality vs. Hype

June 14, 2024

We have always believed that no model of intelligence will ever surpass that of its creator. Despite the advancements in ChatGPT and other generative AIs, this opinion remains unchanged. The clamor for legislation to limit or ban further AI development seems more rooted in corporate politics and territorial ambitions than in actual threats.

These Large Language Models (LLMs) appear impressive and, as demonstrated, can excel in various tests like the SAT and bar exam. However, they perform well on these tests because they have been trained on relevant data. This is akin to a person studying rigorously with the help of tutors and then acing an exam. The irony is that, despite being trained on vast amounts of human-generated content, these models still underperform in other similar evaluations.

Consider the following points to understand our perspective:

Training on Human Content

AI continues to learn from human-generated content. As its use grows, the quality of the content it learns from and produces will decline rapidly. We predict this degradation will occur faster than anticipated, especially for monolithic and consumer-oriented AI models like ChatGPT.

Fair Comparison

For a fair comparison between AI and a human, we should compare a single AI unit running on one machine, disconnected from cloud infrastructure, against one human exposed to the same information (learning dataset).

Human Advancement

Humans are not static. We have our trajectory of increasing knowledge and competence.

Catalyst for Education

AI will likely serve as a catalyst, accelerating personalized education and unleashing human creativity on an unprecedented scale.

Unfair Comparisons

Fears of AI outsmarting humanity often compare an individual against vast computational power with access to all internet content. This is inherently unfair.

Until AI can traverse the journey from Information -> Knowledge -> Experience -> Intuition, there is no need to worry about it taking over. Currently, it is only at the stage of ‘processing information’. Let's observe how quickly it progresses through this journey.

© 2024 ITSoli