AI isn’t human, it’s math
🌈 Abstract
The article discusses the nature of AI, addressing common misconceptions and providing a more nuanced understanding of what AI is and how it works.
🙋 Q&A
[01] AI isn't human, it's math
1. What are the key points made about the nature of AI?
- AI is very broad, with no singular definition, allowing for contradictory claims to coexist
- AI is difficult to detect, with few people aware when they are interacting with it, but it is ubiquitous in everyday life
- Generative AI is more prominent, primarily used by younger, tech-savvy users, while other AI systems are quietly integrated into various applications
- There are many claims and perceptions about AI, ranging from hype to fear, influenced by people's direct experiences with AI products and their knowledge of how AI systems work
2. How is AI fundamentally different from humans?
- AI is not human, but rather data, math, statistics, and complex systems designed to complete specific tasks with variable accuracy
- The quest to replicate human-like intelligence in machines is an aspiration, not how AI actually works
- Humans are incredibly complex, while AI is programmed to complete certain tasks that humans happen to already be doing
3. What are the implications of the human vs. AI comparison?
- It raises questions about the meaning of being human and how we should approach the relationship between humans and AI, whether as partners, competitors, or in terms of AI displacing human jobs
[02] AI is technology
1. What is the history and context of the term "Artificial Intelligence"?
- AI has been around for over 60 years, with the term "Artificial Intelligence" coined in 1956 at the Dartmouth Conference as a way to generate hype and obtain funding, similar to how it is used today
- Every new technology tends to replace human tasks, and the same is true for AI, as it is easier to think of things people already do rather than things people can't do but machines can
2. How should we think about AI as a technology?
- Saying something is an AI product is like saying it is a technology product, as technology and AI are broad categories
- People remember stories, abstractions, metaphors, and products rather than definitions, so the opportunity lies in defining the story of AI
[03] AI is a tool
1. How is AI described as a tool?
- AI is a metaphorical tool, like duct tape, that is good for many things but not everything
- AI can be designed as a tool to help people accomplish certain tasks
2. What are the two perspectives on AI as a tool? a) AI as a design material:
- AI is often already chosen as the underlying mechanism for building an AI product, limiting the ideation process
- Designers then use AI as a material for design, studying its qualities and making usability choices
b) AI as a "hammer searching for nails":
- AI is ready for action but needs to be paired with the right use case for what it can reasonably do
- The match between technology and user need may come later as AI capabilities become possible and reliable
[04] AI is a surrogate of data
1. What is the relationship between AI and data?
- AI systems are as good as the data they are trained on, and their performance has improved significantly due to better data quality and quantity
- Data reflects biases in terms of who collects it, data availability, and data cleanliness, which can impact the performance of AI systems
[05] AI is a partner
1. How is the partnership between humans and AI described?
- AI systems and humans need to understand each other's goals and be willing to intervene and help when needed, creating a true partnership
- This partnership can involve both automation (AI accomplishing tasks on its own) and augmentation (AI assisting humans to accomplish tasks better or in new ways)
- The level of trust and supervision required in the human-AI partnership can vary depending on the situation and task
[06] AI is a new interaction paradigm
1. How does the interaction between humans and AI differ from traditional computer interactions?
- With AI systems, the user no longer tells the computer what to do, but rather tells the computer the desired outcome
- Interaction with AI involves various forms of communication and feedback, not just verbal instructions
- AI enables new types of interactions beyond just language-based communication, such as implicit feedback through actions and behaviors