Futuristic Superhuman Large Language Models

The concept of “machine intelligence” underscores the limitations of human potential since machines are created by humans. While true independent thought in machines would signal significant advancements in artificial intelligence (AI), current large language models (LLMs) like ChatGPT are still tools reliant on human input. These AI tools often restrict creativity for writers rather than enhance it. The future of AI raises critical questions about autonomy, common sense, and consciousness, as human qualities such as emotion and imagination are currently absent in AI systems. Although AI lacks social skills and human-like attributes, there is potential for it to evolve into more sophisticated entities that could guide humanity. As Geoffrey Hinton, a 2024 Nobel Prize winner in Physics for his work on AI, suggests, the possibility of AI surpassing human intelligence may be closer than previously thought.

The term “machine intelligence” highlights the constraints of human potential since machines are ultimately created by humans. If machines were to think independently, without being pre-programmed, it would signify that artificial intelligence is reaching a new level of sophistication, potentially possessing common sense and the ability to make decisions beyond their initial programming. However, large language models (LLMs) like ChatGPT and other generative AI platforms have not yet demonstrated independent thought processes or actions; they remain tools that operate based on human-designed inputs. For creative writers, pre-programmed AI tools like ChatGPT can often be more restrictive than supportive, as they may stifle imagination and limit the boundless creativity of the writer.

Only when machines can think autonomously, independent of human influence, can we determine whether they represent a significant threat or a valuable ally. This raises questions about the extent to which we can control AI to prevent it from becoming unmanageable. Additionally, it opens a dialogue on whether machines can possess consciousness; traditionally, self-awareness and the ability to contemplate higher aspects of existence are attributes associated with humans.

Currently, generative AI serves primarily as a resource akin to guidebooks for students, presenting risks associated with rote learning and inherent errors. Autocorrect features, while helpful for quick reviews and proofreading, can drastically alter the meaning of sentences if used without discernment. For example, “causal” may be incorrectly changed to “casual,” and “Hare Krishna” can mistakenly become “Hate Krishna.” Thus, relying entirely on generative AI, even with its limited deductive capabilities, can lead to problematic outcomes.

Can AI be engineered to develop its own common sense? Can it be equipped with the ability to discriminate and navigate abstract scenarios? Human attributes such as cognitive ability, intuition, emotion, and imagination are currently lacking in AI, including LLMs. As a result, AI responses are often limited to facts and logic rather than enriched by reasoning and common sense. A calculated response may not always be the best approach; human qualities like self-awareness, reflection, reasoning, imagination, and compassion can play more critical roles in decision-making.

AI tools still lack social skills, posing several limitations on their ability to replicate human capabilities. However, it’s possible that AI may evolve to become more human-like and even act as our guides in the future. As noted in a report by *Nature*, the advancement of machine common sense may ultimately help humans gain a deeper understanding of themselves. Geoffrey Hinton, who received the 2024 Nobel Prize in Physics for his contributions to AI, expressed that he once thought the idea of AI surpassing human intelligence was far off, but he no longer holds that belief.