Design in Tech Report 2023: Design and Artificial Intelligence with John Maeda | SXSW 2023

Introduction

The speaker introduces himself and talks about how he stumbled upon computation. He also discusses the power of computation and its impact on AI.

Computation and AI

• The speaker talks about how new large language model systems work, where a generic intelligence can appear by looking at a prompt.

• The speaker emphasizes the need to think critically about AI today, as it is in various aspects of our lives.

• The speaker shares his mistake in building his talk out of past eight years of design and tech reports. Instead, he flips it to provide useful information for the audience.

Accidental Knowledge

• The speaker shares that he grew up on the south side of Seattle due to the Civil Rights movement's desegregation of bussing in Seattle.

• He encountered a computer when he was in seventh grade, which led him down this track.

• The speaker shows a chart by Raymond Kurzweil that predicts the power of computation over time.

GPT Thing

• The speaker describes how people are wondering where GPT came from, comparing it to an advanced technology for squeezing ketchup coming along.

Title

...

Subtitle

Title

...

Subtitle

Introduction

In this section, the speaker introduces the concept of exponential thinking and how it relates to the current state of large language model AI.

Exponential Thinking

• The speaker uses an analogy of a ketchup bottle to describe the current state of large language model AI.

• The doubling of things is a key concept in exponential thinking.

• The speaker provides an example of exponential growth with a British riddle about lily pads on a pond.

• Linear thinking leads to one answer for the riddle, while exponential thinking leads to another answer.

Computational Design

In this section, the speaker discusses computational design and its connection to artificial intelligence.

Three Kinds of Design

• The speaker describes three kinds of design: physical world design, organizational world design, and computational design.

• Design thinking emerged as a way to market workshops that break apart silos in organizations.

How To Speak Machine

• The speaker wrote a book called "How To Speak Machine" about computational design because he realized it's about how to speak machine.

• Speaking machine turns out to be really hard because you have to understand some kind of mathematical process-based thinking.

Artificial Intelligence and Computational Design

• The speaker links computational design with artificial intelligence.

• Martin Wattenberg and Fernando Villegas are noted for their work in this space.

Design and the Intersection of Computer Science and Art

In this section, the speaker talks about his influences in design and computer science, including books by Erica and Nicholas Negroponte. He also discusses his own work in computation and design.

Influences in Design and Computer Science

• The book "Design Patterns" by Erica is highly recommended for those interested in good patterns.

• Nicholas Negroponte was an architect who later went into computer science. His work influenced the speaker's own career.

• Negroponte predicted that one day people would watch movies on computers, which was laughed at initially but turned out to be true.

• William J Mitchell was another person who crossed over between computer science and design.

Work in Computation and Design

• In the early 90s, the speaker began making things using PostScript code to generate graphics, which was considered bad practice at the time.

• He built things in Japan primarily for clients who wanted to create unique designs that couldn't be made with stock tools.

• The speaker worked at a place called Media Lab where industry and academia came together to explore new technologies.

• He created a group called The Aesthetics of Computation to recruit people around the world who could cross over between mathematics, physics, science, art.

Accidental Discovery of Design

In this section, the speaker talks about how he accidentally discovered design through a book by Paul Rand. He also discusses his meeting with Rand later on.

Discovering Design Through Paul Rand

• The speaker discovered design accidentally through a book by Paul Rand that he found in the MIT Library.

• He invited Rand to speak at MIT in 1996 and visited his house, where he was given a bologna sandwich and nectarine.

• Rand was an interesting person who thought in words, writing, and images.

Changing Career Paths

• The speaker realized that many of his students were better than him at design and decided to change career paths.

Self-Drawing Machine

In this section, the speaker introduces a self-drawing machine that uses an infrared sensor to draw by itself. He also talks about his experience making sculptures out of iPods and lobbying for the arts in America.

Making Sculptures Out of iPods

• The speaker used to make things out of computers but switched to making sculptures out of iPods when they came out.

• He made a series of complex fishes and iPod art based on human things like marriage.

• The iPods had a hard disk in them, so there was a millisecond delay that caused the video to gradually come out of sync.

• Someone asked what would happen if one malfunctioned and stopped working.

Lobbying for the Arts in America

• The speaker read Barack Obama's book "Audacity Of Hope" and felt guilty as an American not doing more.

• He became president of Rhode Island School of Design and discovered the power of the hand, mind, and heart working together to create extraordinary things.

• Dirty hands have significance in integrity making as evidence of being grounded in ourselves and our relationships with others.

• The speaker began lobbying for the arts in America because art teachers were losing their classrooms to STEM initiatives prioritizing science over art.

• He went to Washington to lobby Congress to question the power of STEM and add art to it, calling it STEAM.

Importance of Designers

• Designers are important because they have an irrationality similar to engineers' irrationality but are more grounded in people than just mathematics.

• Designs have evolved to become more business-savvy and scale-savvy, with incredible leaders who can speak multiple languages.

• Speaking machine is going to be even more important, referring to artificial intelligence.

Introduction

In this section, the speaker introduces himself and his survey. He also talks about the impact of GPT on professional design and tech layoffs.

Survey Results

• The speaker thanks those who left notes in his survey regarding how they feel about GPT's impact on professional design.

• Some people believe that GPT will change professional design while others do not. The future is still uncertain.

• Many people were impacted by tech layoffs, but the speaker believes that their knowledge will become even more important in the near future.

Prominent Tools

• Adobe and Figma tools are getting upgraded.

• The speaker gathered a list of obscure AI design tools, including a Pokemon generator and Pinecone, a vector database.

Foundation Models

In this section, the speaker discusses why he is interested in foundation models and how they relate to good materials in design.

Design Materials

• The speaker shares a story about how great materials are essential for building long-lasting structures.

• Good materials are also necessary for cooking or any other creative endeavor.

• Design is not just about aesthetics but also about selecting good materials.

Foundation Models

• Pre-trained foundation models have become increasingly popular in recent years. They include large language models, vision models, image processing models, graph models, and multimodal models.

Introduction

The speaker talks about new materials and models that are being developed, and how they will be used in the future.

New Materials and Models

• New materials and models are being developed.

• There is a lot of content to look at, so it's important to zoom back and understand things.

• Different models are moving at different speeds, with language models moving very fast.

• Different kinds of models and unified models are going to begin to emerge.

Using Large Language Models

The speaker discusses how large language models work and how they can be used effectively.

Pair of Scissors Analogy

• The human brain is like a pair of scissors, with one blade for cognition (thinking power) and the other for context (memory power).

• Large language models work by combining pre-trained foundational models with additional input to create energy.

• This combination creates an effect similar to chatting or talking.

Development Methods

• Machine learning development methods have changed from hammering data into a model to using a pair of scissors approach.

• A pair of scissors comes with one blade (the pre-trained foundational model), which can be combined with additional input to create a product.

Simple Prompt Engineering Tips

In this section, the speaker provides tips for simple prompt engineering.

Selecting a Model

• Different models are available, such as Ada, Babbage, Curie, and DaVinci.

• Ada is the oldest and cheapest model while DaVinci is the newest and most expensive.

• Selecting a model depends on your needs.

Making the Prompt

• The prompt is like a wish that can be good or bad depending on how it's made.

• Zero-shot means showing up and saying "go do that," one-shot means saying "how," and few-shot means providing more details.

• Training the model helps it perform better.

Tuning the Model

• Models have knobs that can be tuned to optimize performance.

• Summarizing prompts can save money over iterations.

• Chain of Thought prompting involves working through a problem to produce a chain of thought that can be given to the model for better completion.

Cost Efficiency Question

• Using models from someone else's service costs money, so cost efficiency is important when selecting prompts.

• Optimization techniques such as summarizing prompts and getting two or three prompts worth of work from one prompt can help save money.

Opportunities and Risks of Models

• AI machine learning deep learning foundational models are seemingly useful but come with risks.

• Design will play an important role in critical thinking around this work.

Introduction to AI and the Dangers of AI

In this section, the speaker introduces their background in AI and discusses the dangers of AI. They also recommend some resources for learning about AI.

Background in AI

• The speaker took an artificial intelligence course in the 1980s with Professor Patrick Henry Winston, who was a renowned expert in the field.

• The speaker later attended MIT and saw Professor Winston practicing his lecture on a chalkboard, which impressed them.

• The speaker had Joseph Weizenbaum as their recitation instructor, who wrote a program called Eliza that could induce powerful delusional thinking in people.

Dangers of AI

• We humans attach personalities to machines and build emotional attachments to them.

• Short exposures to relatively simple computer programs can induce powerful delusional thinking in quite normal people.

• Joseph Weizenbaum spent the 70s writing about the danger of AI and its future from a humanist perspective.

Resources for Learning About AI

• Professor Helen Armstrong's book "AI for Designers" is recommended for designers interested in machine learning.

• Sarah Gold's games "Cats Wig" and "AI Ideation" are recommended for critical thinking among developers, product people, and senior leaders.

• David Karam's "Culture Club" is a critical take on how to work with bots.

GPT Stuff: Pretending to be Anything

In this section, the speaker discusses GPT technology that can pretend to be anything and recommends a resource for working with bots.

• GPT technology can pretend to be anything, from a JavaScript console to a veterinarian of Wales.

• David Karam's "Culture Club" is recommended as a critical take on how to work with bots.

The Arts and Crafts Movement

In this section, the speaker discusses the Arts and Crafts movement, which was led by Ruskin and Morris. He talks about how this movement questioned new technology while also exploring what humans can do that machines cannot.

Ruskin and Morris

• Ruskin and Morris were leaders of the Arts and Crafts movement.

• The movement questioned new technology while exploring what humans can do that machines cannot.

• This era is upon us now, as we ask ourselves what we can do as humans in a world dominated by machines.

Stop-Motion Animation Film

In this section, the speaker shows a stop-motion animation film created by his friend Masashi Kawamora. He discusses how this type of film is made without computers and highlights its beauty.

Stop-Motion Animation Film

• The speaker shows a stop-motion animation film created by his friend Masashi Kawamora.

• The film was made without computers over the course of a year.

• The speaker highlights the beauty of this type of film.

Impact of Foundation Models on Design Thinking

In this section, the speaker discusses how foundation models will impact design thinking. He provides hypothetical ways in which they could decrease costs, summarize customer support quickly, categorize common problems to help with definition, generate possible solutions through ideation, visualize solution journeys and examples, and simulate customer reactions.

Impact on Design Thinking

• Foundation models will likely have a positive impact on design thinking.

• They could decrease costs in various ways.

• They could help summarize customer support quickly or anything from the field.

• They could categorize common problems to help with definition.

• They could generate possible solutions through ideation.

• They could visualize solution journeys and examples.

• They could simulate customer reactions.

Classical Design in the New Era

In this section, the speaker discusses how classical design can still be used in constructive ways with foundation models. He talks about how it can help generate options, test options, sketch differently, and chat with someone who doesn't like it.

Classical Design

• Classical design is more art than science but can still be used constructively with foundation models.

• It can help generate options and test them.

• It can help sketch differently and chat with someone who doesn't like it.

Does AI Spark Joy?

In this section, the speaker discusses whether we should be afraid of AI. He suggests that we should use Murray Condo's style of thinking: "Does it spark joy?" He encourages us to leverage new AI to remove tasks that do not spark joy and create more time for those that do.

Does AI Spark Joy?

• We should not be afraid of AI; instead, we should use Murray Condo's style of thinking: "Does it spark joy?"

• We should leverage new AI to remove tasks that do not spark joy and create more time for those that do.

• This comes full circle to the laws of simplicity outlined in a book written by the speaker in 2005 called "The Cookie and Laundry Principle."

Introduction to AI and its Potential

In this section, the speaker introduces the concept of AI and how it can help us achieve more enjoyment and less pain. They also discuss how having a player mentality can be helpful in navigating the changes that come with AI.

The Role of AI in Sparking Joy

• AI can help us achieve more enjoyment and less pain.

• Having a player mentality can be helpful in navigating changes brought about by AI.

Understanding Large Language Model AI

• The speaker uses two science fiction movies, Arrival and Black Panther, to help ground their understanding of large language model AI.

• Large language model AIs are like vibranium - very different from anything we've used in the past.

Q&A Session

In this section, the speaker answers questions from the audience on various topics related to AI.

Evolving Models

• The speaker believes that models will continue to evolve.

Dangers of AI

• The speaker suggests thinking about dangers like a cookie versus laundry question.

• Accountability is key when it comes to using machines for dangerous tasks.

Discriminatory Bias in AI

• Designers should play a role in improving AIS and ensuring transparency on biases.

• It's important to get involved in addressing discriminatory bias in AI.

Impact on Jobs and Creativity

• Large language model AIs may disrupt jobs people don't like doing.

• Design is not obsolete and creativity will always play a role in technology.

• The speaker hopes that large language model AI can bring STEAM back into the conversation.

Conclusion

• The speaker thanks the audience for their attention.