* This blog post is a summary of this video.

The Rise of AI: Understanding ChatGPT and The Future of Artificial Intelligence

Author: LastWeekTonightTime: 2024-01-23 06:25:00

Table of Contents

What is AI and How is it Used in Our Daily Lives?

Artificial intelligence (AI) is increasingly becoming a part of modern life, from self-driving cars to spam filters to creepy robot therapists. While it may seem like everyone is suddenly talking about AI, the truth is it has already become embedded in many aspects of daily life, often without people even realizing it. For example, your phone likely uses AI for facial recognition to unlock it or predictive text suggestions. Smart TVs also utilize AI to recommend content or adjust picture quality.

AI is also already being used by companies to make critical decisions, like determining whether or not a job applicant even gets seen by human eyes. The CEO of ZipRecruiter estimates that 75% of resumes submitted in the US are first scanned by an algorithm, which decides if they warrant further review. So AI is already deeply impacting areas like employment, often without transparency into how these systems work.

AI on Your Phone and TV

Your phone uses AI for features like facial recognition to unlock it, predictive text suggestions when typing, and organizing photos. Smart TVs also use AI to recommend content to watch or automatically adjust picture settings for the best quality. So while you may not think about it, AI has become deeply integrated into common devices, enhancing certain functionality in helpful but also potentially concerning ways when it comes to privacy issues.

AI Making Critical Decisions

Companies increasingly use AI systems to scan and filter resumes, essentially deciding whether a human ever sees a particular application. Over 3/4 of resumes submitted in the US go through such algorithms first. Other examples involve AI programs being developed to detect certain medical conditions earlier and more accurately than doctors, which could significantly improve and even save lives. But flawed data or algorithms could also lead to missed diagnoses and lack of explainability into the system's determinations.

Narrow vs. General AI - Understanding the Key Differences

When discussing artificial intelligence, it's important to understand the difference between narrow AI and general AI. Narrow AI systems can perform only narrowly defined tasks, while general AI would demonstrate intelligent behavior across a wide range of cognitive functions.

The AI currently in use, including the image generators and chatbots that have caused recent excitement, are narrow AI. They can generate human-like writing or artwork around a specific prompt, but cannot match the versatility of the human mind. True general AI does not yet exist and remains theoretical.

How Deep Learning has Led to Major Advances in Narrow AI

Recent exponential improvements in narrow AI systems are thanks to a method known as deep learning. Instead of detailed human programming, these systems are given minimal instructions but massive amounts of data from which they essentially teach themselves.

For example, when researchers gave a deep learning system the basic goal of winning at the Atari game Breakout, within 500 games it had taught itself to innovatively dig a tunnel allowing the ball to bounce around bricks. This ability to interpret data and improve with experience explains the new creative power of tools like image generators and chatbots.

Exciting Potential Applications of AI in Areas Like Medicine

There are many promising uses of AI in development, especially in the medical field. Systems are being trained to listen to voice pattern changes to detect conditions like Parkinson's early or analyze protein shapes to accelerate drug discovery. Experts say AI could 'change medicine' and greatly improve treatment.

But there are risks if the underlying data or algorithms have flaws and biases. Joy Buolamwini found that some facial tracking tools performed worse on darker skin tones, likely due to insufficient diversity in the training data. So while the potential is great, equitable access and outcomes are not guaranteed.

Concerns Over AI's Impact on Jobs and The Economy

With AI able to replicate certain tasks better and faster than humans, there are understandable economic concerns about it replacing human jobs. However, experts point out that previous bouts of automation primarily affected blue collar jobs, whereas AI may functionally change white collar jobs instead.

Just as past technological shifts ended some jobs but created new ones, AI will likely continue that pattern. Most expect a restructuring rather than wiping out of roles like lawyers, who could use AI tools to become more efficient. But there will undoubtedly be workforce disruptions along the way that must be accounted for.

Troubling Ethical Issues Around AI and Art

The use of massive datasets to train AI art generators has raised accusations of plagiarism, since systems scrape images without artist permission. Getty Images is even suing one company whose tool produced a distorted image still bearing their logo.

This speaks to wider concerns about properly crediting sources and ensuring artists maintain control and rights over their creations versus AI systems trained on appropriated data. Generative programs also threaten the livelihood of artists whose style and work it imitates for free.

The Black Box Problem - The Lack of Explainability in AI

A major issue with AI systems right now is something experts call the 'black box problem.' When an algorithm teaches itself from data and the final model is too complex for even its creators to fully comprehend, it becomes impossible to explain exactly how or why it arrived at a particular decision.

This lack of transparency is why problematic biases can emerge unchecked in areas like hiring software. It also means when AI chatbots declare something nonsensical like loving a reporter, the companies themselves admit they cannot determine why it happened or was generated.


Q: What is the difference between narrow and general AI?
A: Narrow AI can only perform specific predefined tasks while general AI would be able to demonstrate intelligence across a wide range of cognitive functions, similar to human intelligence.

Q: How has deep learning enabled advances in narrow AI?
A: Deep learning allows AI systems to train themselves by ingesting large datasets rather than needing explicit human programming, allowing for exponential increases in capability.

Q: What is the black box problem in AI?
A: The black box problem refers to the opacity of many AI systems, where even their programmers cannot explain the reasoning behind results, posing ethical issues.