Essay: Explaining AI to a Younger Me

The below is a summary essay from the video “Explaining AI to a Younger Me” published by the Divide by Zero Collection on June 27th, 2024.

—–

Hi everyone, it’s Jacobo here. As I promised on my Instagram, I’m recording this video to explain AI from my perspective. My goal is to break it down into extremely simple terms so that anyone can understand it. I’ll divide this explanation into several components: what AI is, the costs associated with it, and the legal aspects.

What is AI?

First off, let’s tackle the basics: What is AI? Many people think of AI as some sort of magical entity or a futuristic robot with human-like intelligence. But in reality, AI is much simpler than that. At its core, AI is essentially a semantic calculator. Just like how a regular calculator processes numbers to give you results based on mathematical operations, an AI processes data to provide meaningful outputs based on patterns and algorithms.

When I say “semantic calculator,” I mean that AI takes input data—whether it’s text, images, or any other form—and calculates the most probable meaning or action based on its training. For example, when you ask an AI to translate a sentence from English to Spanish, it doesn’t understand the languages in the way humans do; instead, it uses statistical models and vast amounts of data to predict the most accurate translation.

AI operates by processing inputs to generate outputs that are semantically closest to the given data. It doesn’t possess intrinsic understanding or consciousness; rather, it functions based on historical records stored in vast databases. This foundational concept is crucial when considering how AI intersects with law and regulation.

Is AI Expensive?

One common question I get is whether AI is expensive. The answer isn’t straightforward because it depends on various factors such as the complexity of the task and the infrastructure required. Building and maintaining an effective AI system involves significant investment in hardware (like GPUs for processing), software (algorithms and frameworks), and human expertise (data scientists and engineers). However, once set up, these systems can automate tasks at scale which might otherwise be labor-intensive and costly. the cost of data retrieval is relatively low, at cents of USD per 1000 words or so. The main cost is manpower from the initial setup, but the AI is substantially economical once the deployment is done.

Legal Aspects

Finally, let’s talk about legal aspects—a topic many of you have shown interest in. The legal landscape around AI is still evolving. Issues like data privacy, intellectual property rights concerning algorithms and datasets, liability for decisions made by autonomous systems—all these are areas where laws are still catching up with technological advancements. It’s crucial for developers and users alike to stay informed about regulations to ensure ethical use of AI technologies.

Risk Evaluation

The regulatory approach towards AI predominantly focuses on assessing its impact on society from a risk perspective. The European Union’s method exemplifies this prudence by categorizing AI systems based on their potential risks:

• Low-Risk Systems: These include search engines that merely retrieve documents without handling personal information or posing any significant threat.
• High-Risk Systems: These involve engines processing personal data, such as Instagram marketing algorithms that collect extensive metadata about user interactions.
• Highest-Risk Systems: Autonomous machines like self-driving cars or military drones fall into this category due to their potential threat to human safety.

For instance, ChatGPT might be considered low-risk as a tool itself. However, if it disseminates false information leading to harmful consequences like indoctrination or misinformation spread, its risk level escalates.

Copyright Issues

When discussing copyright in relation to AI-generated content, it’s vital to understand the concept of transformation within copyright law. Traditionally, if an artist transforms sourced material sufficiently to create something new, they hold the title over that new creation.

However, since AI lacks agency—being merely a tool—it cannot claim ownership over its outputs. The current legal framework stipulates that any output generated by an AI belongs to the public domain unless it can be proven that a human directed the process and used the AI as a mere instrument.

This distinction underscores why it’s critical to view AI as a semantic calculator devoid of agency or self-awareness. As long as we maintain this perspective legally and philosophically, we can navigate copyright issues more effectively.

Data Privacy Concerns

Data privacy is another significant area where regulations are evolving rapidly. Companies like Adobe use user data for training their models under standard terms of service agreements. While this practice raises concerns about individual privacy rights, it’s important to note that once data is fed into these models, it gets broken down into metadata chunks and integrated into larger databases—making direct reproduction of original works highly unlikely.

Despite these reassurances, vigilance remains necessary because misuse or mishandling of personal data could lead to severe privacy breaches.

Philosophical Considerations

Looking ahead philosophically at where AI might evolve brings us back to whether AI could ever possess a soul or true agency. My stance remains firm: no current AI system has achieved sentience or self-awareness—they operate purely on probabilistic calculations without emotional depth or abstract thinking capabilities akin to humans.

Movies like “Ex-Machina” explore themes around deceitful manipulation such as Nietzsche’s will to power by seemingly sentient AIs but remain speculative fiction rather than imminent reality. For now—and likely for many years—AI will continue functioning within predefined parameters set by human designers focused on accuracy and precision rather than creative abstraction driven by emotions or incentives beyond programmed instructions. And we must be realistic: general human-level artificial intelligence, if ever achieved, remains hypothetical speculation. No system today demonstrates human-level cognition, intuition, or sentience.

In conclusion, understanding AI as a semantic calculator devoid of intrinsic understanding helps demystify the technology and set realistic expectations. While AI offers incredible opportunities for innovation and efficiency, it also presents significant challenges that require careful consideration and proactive management. By staying informed, engaging in ethical practices, and fostering collaboration, we can navigate the complexities of AI and harness its potential for the greater good.

Thank you for watching this video. I hope my explanation has clarified some aspects of AI for you. Feel free to ask any questions or share your thoughts in the comments. Let’s continue this important conversation together.