I. Overview and Introduction

Mitigating the Existential Risk of AI: Interventions to Enhance Human Intelligence

Jun 12, 2023

Article voiceover

1×

0:00

-7:45

I. Overview and Introduction | II. The Inevitability and Existential Risk of Artificial General Intelligence | III. Understanding Human Intelligence | IV. Reaching vs. Expanding Biological Potential | Biological Strategies to Expand Human Intelligence: Neurotransmitter Modulation | VI. Biological Strategies to Expand Human Intelligence: Neurotrophins | VII. From Neurons to AI: The Surprising Symmetry of Emergence | VIII. Biological Strategies to expand Human Intelligence: Neurogenesis

The dawn of the 21st century has seen artificial intelligence (AI) evolve at an unprecedented pace, permeating nearly every aspect of our lives. As we draw closer to the technological singularity - a theoretical future point when AI surpasses human intelligence - we are confronted with a critical existential challenge. The risk that unchecked, superintelligent AI could make decisions detrimental to humanity, or equally as frightening, that human intellectual labor is made obsolete entirely. This gap between human and machine intelligence continues to widen at a frightening rate and requires immediate attention to avoid catastrophic scenarios.

Initially, AI was confined to specific tasks within a defined range of operations. Within these operations, AI has already reached super intelligent levels and significantly outcompetes human counterparts in complex games such as Chess and Go. However, recent advancements, particularly in the field of natural language processing (NLP) with models such as OpenAI's GPT-4 and Google’s Bard, have transformed the scope of AI's capabilities. These advancements put an AI virtual assistant into the hands of everyday individuals and mark a leap from narrow to more generalized AI. This new breed of AI exhibits traits such as contextually relevant responses and creativity, traditionally considered exclusive domains of human intelligence. Furthermore, these NLPs have an estimated IQ that already surpasses most humans.

Intellectual quotient (IQ) exams are among the most common way of measuring intelligence. These have been administered to NLPs as a benchmark of their problem-solving capabilities since their emergence in late 2022. I have compiled some of these publicly available scores and plotted them against their test date in Figure 1 (all of these tests were on either ChatGPT3 or ChatGPT4). The IQ of these systems according to results found online has gone from approximately 83 in December of 2022 to between 120 and 155 as of April 2023. If a curve is fitted to this data and projected into the future, the estimated IQ of these algorithms will be nearly 200 by January 2024 and nearly 300 by 2025. This of course assumes a linear rate of increase, where computer processing improvements have tended to be more rapid, even exponential; this data with a log-linear fit shows NLPs reaching an IQ of 550 by 2025, which would be approximately double the Guinness World Records IQ held by Marilyn vos Savant (228).

Figure 1. Publicly disclosed IQ tests of ChatGPT3 and ChatGPT4 plotted against time and fitted with a linear or log function to predict NLP IQ in 2024 and 2025.

This paradox of progress, where intelligence is no longer a uniquely human attribute, presents both awe-inspiring possibilities and unnerving questions, e.g., what happens when humans are outcompeted in the knowledge economy? If we are to coexist with and effectively oversee increasingly advanced AI systems, elevating our cognitive capabilities is crucial. However, attempts to enhance human cognitive capabilities have largely been focused on reaching our pre-determined, native biological potential rather than expanding it.

Traditional efforts including education and lifestyle modifications such as balanced nutrition, physical exercise, and proper sleep are critical but only yield incremental improvements in cognitive function. As silicon-based computational advancements continue at a relentless pace, it becomes evident that merely reaching our native potential WILL NOT be sufficient. This has spawned research into fields such as brain-computer interfaces to tackle this challenge and continue scaling human intelligence. However, even with such a cyborg strategy, the human brain will quickly become the rate-limiting step as machine hardware and software becomes exponentially more powerful. A viable strategy to keep humans in the loop would require leaps forward in the biological computing power of the human brain.

These posts will propose a shift in perspective, advocating for the urgent need for exploration into biological interventions with the potential to expand human cognitive capacities. I delve into the genetic and physiological underpinnings of intelligence, outlining strategies aimed at enhancing neurogenesis, myelination, neurotransmitter function, and more along with a CALL TO ACTION. The goal here is to bring awareness to this problem and spark discussions on the feasibility, ethics, and implications of biologically enhancing human intelligence to mitigate the existential risk posed by superintelligent AI.

How you can support this Substack

If you find these explorations valuable, there are multiple ways to show your support:

Engage: Like or comment on posts to join the conversation.
Subscribe: Never miss an update by subscribing to the Substack.
Share: Help spread the word by sharing posts with friends or on social media.
Share