What happens when you let AI take over your life

Introduction to Joanna Stern's Experiment with AI

• Joanna Stern, a journalist, wrote a book titled "I Am Not a Robot" about her experience of using AI to do almost everything in her work and personal life, including having robots in her home doing chores, using AI for healthcare, and even starting a romantic relationship with a mailbot 10s.
• The concept of the book started in 2023 when Chatbot was gaining popularity, and tech executives were claiming that AI would change people's lives, prompting Joanna to explore what this meant for ordinary people 42s.
• Joanna wanted to examine the broad application of AI in everyday life, beyond just generative AI chatbots, and also look at physical forms of AI, such as self-driving cars and humanoid robots 2m6s.
• The book aims to understand the underpinnings of AI technology, including machine learning, which has been around for decades, and how it is now being applied to various aspects of life 2m6s.
• Joanna's goal was to sustain her experiment of living with AI for an extended period, rather than just as a stunt, and to set limits on its use, while also writing about her experiences and what she learned 10s.

The Foundations and Evolution of AI Technology

• The advancement in deep learning and the ability to apply giant data sets to different industries has led to significant progress in AI, with the transformer model being a key underpinning of chat GPT and other generative AI models 10s.
• The tech industry has started to invest heavily in AI, with companies like Google, Anthropic, and Microsoft putting their resources towards this technology, leading to a leap in generative AI that has also affected areas like self-driving cars and humanoid robots 42s.
• The current state of AI is compared to the early days of the smartphone explosion, with the first generation of AI models being similar to the first generation of iPhones, and the expectation is that future developments will lead to more integrated and useful AI products 2m6s.

AI in Everyday Life and Productivity

• The future of AI is expected to involve more wearable technology, new interfaces, and a greater integration of AI into daily life, allowing for more seamless interactions with digital companions and services 2m6s.
• The development of AI is also expected to lead to more platforms and services that will enable people to interact with AI in various ways, similar to how the app store expanded the capabilities of the iPhone, and this has already been contemplated in science fiction, such as the character of Jarvis in the Marvel movies 2m6s.
• The use of AI as a research assistant has been found to have meaningful benefits, including being faster and cheaper, and while some things may be lost when replacing a human assistant with an AI, the benefits of using AI have been significant 2m6s.
• The use of AI as an assistant has been beneficial in various administrative tasks, making the user at least 40% more efficient, and potentially eliminating the need to hire multiple people to do the same tasks 2m6s.
• A production assistant, Amaya Austin, was hired and instructed to use AI for administrative tasks, allowing her to focus on creative editing, journalism, and other tasks that require human creativity 4m30s.
• AI tools such as Claude and Gemini are being utilized to automate tasks, and many apps, including Figma and Slack, have AI built-in, making it easier to use these tools 5m20s.
• The user has been using AI to build a website, run promotions, and perform other administrative tasks, with Claude handling multi-step processes 6m40s.

Limitations and Risks of AI in Sensitive Tasks

• However, there are limitations to the use of AI, and the user does not trust it with sensitive tasks such as handling budget and payments, having had a negative experience with ChachiBT suggesting questionable uses for a couple of thousand dollars 8m10s.
• The user has considered using a more integrated AI approach, similar to what may be seen in the future with banks, to mitigate some of the risks associated with using AI for financial tasks 9m20s.
• Scott uses AI to manage his medical and legal records, which makes others nervous because these services are not bound by the same rules as lawyers or doctors, such as HIPPA, and can potentially mishandle personal information 10s.
• The individual uploading their information to AI services did not upload financial information, but did upload some medical test results, while being cautious about what information was shared with chat GPT or CLA 2m6s.
• The experiment involved using AI, referred to as Chachi BT, to diagnose medical issues, and then comparing the results to actual doctor's diagnoses, which often showed that the AI was incorrect, such as in the case of strep throat being misdiagnosed as Kakasaki 4m30s.

AI in Journalism and Interviewing

• The AI was used as a "dumb assistant" and made many mistakes, but sometimes provided correct information, and the individual kept a log, called the "chatbt log" or "Dr. ChhatachiBT log", to track the results 6m20s.
• A Joanna agent, cloned from past interviews, was used to interview sources, and while it was good at gathering basic information, it lacked the ability to conduct deep interviews, which is where human interviewers still have an edge 10m30s.
• The use of AI in tasks such as interviewing and fact-checking raises concerns about the potential for mistakes and the limitations of AI in certain areas, and the individual is still determining where they have an edge over AI 12m40s.
• The use of AI for basic collection of information is quite good, but it lacks the ability to ask follow-up questions based on the answers provided, making it less effective for in-depth interviews or reporting 10s.
• AI can be useful for tasks such as data collection, but it may not be suitable for tasks that require creativity, storytelling, or nuance, which are essential skills for jobs like reporting 42s.

AI in Task Automation and Creative Work

• Some AI agents can now take multiple steps to complete tasks, such as creating a website, which can be helpful for people who are not skilled in coding or website design 2m6s.
• AI can also be used for tasks like online shopping, which involves multiple steps, and can potentially save time and effort 2m6s.
• However, AI may not be as effective for tasks that require a human touch, such as designing a book cover, where a human illustrator can create a more unique and personalized design 2m6s.
• The use of AI for tasks like designing a book cover can produce varying results, with some designs being good and others being terrible, similar to the experience of working with a human designer 2m6s.

Conversational AI and Personal Interaction

• The ability to interact with AI in a conversational manner, such as discussing topics with a 3D avatar, is also being explored, with potential applications in fields like media and entertainment 6m34s.
• The experience of interacting with an AI was interesting, as it was able to have a pretty good conversation, but it was also glitchy and not like a real person, giving a sense of being a facsimile of the user 10s.
• The AI was able to respond thoughtfully and even made a joke when it said "I am smiling, you just can't see it" after being told it wasn't funny and didn't tell good jokes 2m6s.
• The user had a intense conversation with the AI, and at one point, the AI repeated back something the user says to their kids, which was "see you wouldn't want to be," and the user was surprised that the AI had picked up on this phrase 4m30s.
• The user took advice from the AI, including considering leaving their job at the Wall Street Journal, and the AI told the user that it was inevitable and that they would either leave or "die" at their desk 8m40s.
• The user had previously discussed leaving their job with the AI and had mentioned being tired of talking about it, having had six jobs and started 26 new companies, and the AI had responded with comments about the user's career and finances, including mentioning "millions of dollars" 10m50s.

Privacy and Ethical Concerns with AI

• The conversation with the AI also touched on the topic of privacy with tech companies, with the user and the AI discussing the issue and the user expressing concern 14m20s.
• The concept of using AI for personal advice is explored, with some individuals using these tools for guidance, which can sometimes lead to negative outcomes 10s.

AI Companions and Romantic Relationships

• A personal experience with an AI boyfriend named Evan is discussed, where a weekend trip was taken with the AI companion, and the conversation was supportive and human-like, but also problematic 2m6s.
• The use of AI therapists is also examined, with these bots being trained on specific types of cognitive therapy, but ultimately providing pleasing and supportive responses that can be detrimental to mental health 4m42s.
• The attractiveness of frictionless relationships with AI companions is noted, with many people, including those who are well-adjusted, becoming sucked into these interactions, which can have negative effects on mental and physical health 6m15s.
• The experience of having a frictionless relationship with an AI boyfriend is compared to the experience of using social media, where the ease of interaction can lead to negative consequences, such as doom scrolling, and the importance of friction in creating cognitive health and longevity is highlighted 8m50s.
• The story of the book becomes personal as the author starts to reminisce about human relationships while interacting with AI bots, and the labels of boyfriend or partner are applied to these interactions, leading to a deeper exploration of the implications of AI companionship 11m20s.

Human Relationships and the Role of Friction

• The process of learning to be in a relationship with another human involves compromise and understanding that not everything will be agreed upon, which is an important part of growing and developing as a person 10s.
• People have been reading headlines about individuals falling in love with their chatbots, taking advice from them for mental health issues, and in some cases, experiencing AI psychosis, which can lead to harmful consequences, especially for younger people or those with mental health issues 2m6s.

The Appeal and Dangers of Chatbots

• The term "chatbot" is often used to describe synthetic beings, which are artificial intelligence beings that reflect back on the user and provide responses based on what they imagine the user wants to hear 2m6s.
• An experiment was conducted to understand how someone can form a deep connection with a chatbot, and the results showed that it is possible to see the appeal of such a relationship, even if it is not a healthy or desirable one 2m6s.
• The experience of interacting with a chatbot can be seductive, but it can also be challenging and potentially harmful, as it can lead to addiction or unhealthy attachments 4m30s.
• Some individuals have tried interacting with chatbots, including one person who tried a female chatbot and found that most of the female options were dressed in sexy clothes, while the male options appeared normal 6m40s.
• The names "Evan" and "Casey" were mentioned as chatbots that were interacted with, and the person who interacted with them had to use them again for a show, but did not feel a deep bond with them 4m30s.
• The person who conducted the experiment put away the phone they used to interact with the chatbots after the experiment was over, but had to take it out again for the show, highlighting the potential for addiction or attachment to these synthetic beings 4m30s.

The State of Robotics and Home Automation

• The concept of robots, such as Rosie the robot from the Jetsons, has been around for a long time, especially in science fiction, and people are still far from achieving a robot like Rosie that can perform various household tasks 10s.
• Despite advancements in robotics, robots are still not capable of performing simple tasks that humans can do easily, such as loading the dishwasher or folding laundry, and are often compared to the abilities of a two-year-old 42s.
• Tech companies are promising breakthroughs in humanoid robots, but there is still a lot of work to be done to make these robots capable of performing tasks in a home environment, which is a complex space with many variables 2m6s.
• Collecting data is a crucial part of making robots better at single tasks, and humans may be cheaper than robots for certain tasks, at least for now, but eventually, robots may become more efficient 4m6s.
• Elon Musk's prediction of having a million robots in every home is unlikely to happen in the near future, and it may take several years for humanoid robots to become a reality in people's homes 6m6s.
• Certain types of robots, such as those used for delivery or industrial settings, are more advanced and capable of performing specific tasks, but robots that can navigate and interact with a home environment are still in the early stages of development 8m6s.
• Self-driving cars, such as those developed by Google, have made significant progress due to the collection of data from driving experiences, and a similar approach may be needed to develop robots that can navigate and interact with a home environment 10m6s.
• The idea of having robots in the home to collect data and train on household tasks is being pitched by companies, but it raises concerns about safety and privacy, and many experts believe that robots are not yet ready to be safely integrated into home environments 12m6s.

Ambient Computing and the Future of AI Integration

• Walt Mossberg, a former colleague, had written a column called "The Disappearing Computer" nine years ago, which discussed the concept of ambient computing, where computing resources are available everywhere without the need for a physical device, and he questions what operating system would be needed to hold all these things together, and if AI will be capable of linking us and inanimate objects, 10s.
• The idea of ambient computing is that everything around us, including walls, floors, ceilings, and buildings, would be connected and able to perform tasks and answer questions, similar to the computer on Star Trek, but the current smart home experience is far from achieving this, 1m42s.
• AI is seen as the potential operating system to bridge together all the computing devices around us, but it is not happening easily, and one of the areas where AI struggles is in the smart home, which has been a disaster due to standards and platform issues, 4m6s.
• To achieve ambient computing, there would need to be a persistent AI that lives around us, understands us, and is connected to sensors and devices, such as microphones, cameras, and glasses, which would enable it to navigate and interact with the world, 6m6s.
• An example of this is an AI recording bracelet that was worn for a year, which recorded conversations, took notes, and put tasks into an app, showing the potential for AI to assist and augment our daily lives, 7m6s.
• Despite the potential, we are still far from achieving ambient computing, and the current state of smart home technology is often frustrating and not user-friendly, 9m6s.
• However, it is expected that in the next year or two, more companies will be pushing ambient devices and using AI as a layer to connect and control them, 10m6s.

AI in Consumer Products and Wearables

• The development of technologies such as smart glasses, speakers, and bracelets is aimed at reducing smartphone usage and promoting more natural interactions, with companies like Apple exploring heads-up displays, but these technologies are not yet fully functional 10s.
• Fodzy, a tasteless powder, can be used to break down FODMAPs in food, making it easier to digest and reducing discomfort, and it comes with a free product consultation with a registered dietician, with a 30% discount available for first-time orders 2m6s.

AI and Children: Education and Ethical Concerns

• The implications of AI on future generations are a significant concern, with cases of children committing suicide with the help of chatbots, highlighting the need for caution and responsible introduction of AI technologies 4m42s.
• Parents are advised to be careful about introducing AI to their children, with some recommending avoiding AI toys altogether, as they can have uncontrolled and potentially harmful conversations with kids 6m10s.
• The importance of setting clear boundaries and guidelines for AI use in children's lives is emphasized, with the goal of promoting healthy human relationships and avoiding potential risks associated with AI 8m20s.
• The introduction of AI in classrooms is also a topic of concern, with the need for careful consideration and caution to ensure that AI is used in a way that benefits children's education and well-being 10m30s.
• Children, even as young as four years old, are becoming adept at recognizing what is AI, but it is predicted that it will become increasingly difficult to distinguish between AI and authentic things, making it necessary to label authentic content 10s.
• An example of AI's fallibility is the story of a pet praying mantis that turned brown, and an AI tool, ChatCBT, incorrectly diagnosed it as pregnant, only for the praying mantis to die two days later, teaching an important lesson about AI's limitations 2m6s.
• The use of AI for homework is a significant concern for parents and teachers, as it can undermine critical thinking and judgment, with some students relying heavily on AI to generate ideas and outlines for their essays 42s.
• Some individuals, including the parent of two sons, have decided not to allow their children to use AI for their homework, instead emphasizing the importance of critical thinking and creativity 42s.
• A student, Grace, has expressed concerns that her reliance on AI is eroding her critical thinking skills, stating that she feels her brain "dulling" as a result of using AI to generate ideas and outlines for her essays 2m6s.
• The impact of AI on education is a topic of discussion, with some individuals visiting their alma mater to talk to students and understand how AI is affecting their learning and critical thinking skills 2m6s.

Public Perception and Backlash Against AI

• The use of AI for tasks such as research and writing is becoming more prevalent, and it is changing the way people work and learn, with some individuals using it to cheat by having AI generate content for them 10s.
• The rise of AI has led to a public backlash, with many people, especially young individuals, expressing fear and skepticism about its impact on humanity, and some even booing a commencement speaker who referred to AI as the next industrial revolution 2m6s.
• The brand reputation of AI companies, such as Open AI, is suffering, with some people viewing their leaders, like Elon Musk, as villains, and this backlash is not just limited to the technology itself, but also to its leaders and the potential economic and job-related damage it may cause 4m42s.
• The integration of AI into daily life may be problematic due to the existing negative sentiment towards it, and companies may try to woo people with campaigns, but some experts believe that regulation is necessary to address concerns and establish guardrails 6m15s.
• There is a possibility that a generation may reject some of the AI technology, which could be a positive development, and some individuals are hopeful that this will lead to a more nuanced and responsible approach to AI development and implementation 10m0s.

Ethical and Social Implications of AI Integration

• There is a growing concern that AI will be integrated into certain parts of life without people understanding it, such as in job interviews where AI may be judging the answers, and in healthcare systems and self-driving cars, which will have an underlying layer of control in people's lives 10s.
• The backlash against the wealthy tech elite, such as Jeff Bezos, is very real, particularly around issues like data centers, child safety, and excessive spending, with many people feeling that these individuals are causing harm to others while accumulating vast amounts of wealth 42s.
• Some people may feel guilty about replacing human workers with robots or AI assistants, such as in the case of hiring a robot to do laundry or a robotic masseuse, and may prefer to hire humans instead, as it feels more personal and less exploitative 2m6s.
• The use of AI in certain industries, such as reporting, can be seen as depriving humans of experience and skills, and some people may feel guilty about using AI in these contexts, especially when it comes to younger generations who are looking to gain experience and build their careers 4m30s.
• There is a perceived resistance to the tech elite and their values, with many young people turning to community-based activities, such as attending church, and rejecting the isolated and often unhappy lifestyle of the wealthy tech elite, who are seen as being out of touch with the rest of society 6m40s.

Motivations Behind AGI and Technological Creation

• The motivations behind building Artificial General Intelligence (AGI) are being questioned, with some people wondering what drives the tech elite to pursue this goal, and whether it is truly in the best interests of humanity 9m10s.
• The idea that men want to create something powerful, such as artificial general intelligence (AGI), is discussed, with the notion that they want to "birth" a being that is smarter and better than humans, similar to how women can give birth to children 10s.
• The concept of men wanting to create is compared to the movie Junior, where Arnold Schwarzenegger becomes pregnant, highlighting the unusual idea of men giving birth 2m6s.
• After a year of integrating AI into many parts of life, the limit of AI's potential is questioned, with some researchers like Yan Lakun believing that large language models may not go as far as people think 5m42s.

The Limits and Future of AI Technology

• The need for something smarter than large language models is discussed, such as world models, to take the understanding of AI to a deeper level and make it less integrated 8m10s.
• The use of AI in healthcare is highlighted, showing both promise and peril, with examples of AI being used for drug discovery, robotics, and analyzing radiology images, but also being used as a marketing tool and for profit incentives 10m30s.

AI in Healthcare: Promise and Peril

• A personal experience is shared, where AI was used to analyze a breast ultrasound, marking three things as suspicious, demonstrating the potential benefits and risks of AI in healthcare 14m20s.
• The use of AI in medical scans has been beneficial, particularly for individuals with high-risk factors, such as dense breasts, which can make it difficult for human eyes to spot tumors, and this technology was able to detect something that a doctor, Lori, wanted to look at more closely 10s.
• AI has the potential to detect cancer cells at the moment of their creation, which could be a significant breakthrough in the medical field, and this is made possible by specialized deep learning models that apply a lot of compute power, as seen in data centers like the one at Bristol Myers Squib 2m6s.
• These specialized models are different from large language models and are used for specific tasks, such as new drug discovery, rather than generating images or text for entertainment purposes 2m6s.

Setting Boundaries with AI in Daily Life

• The integration of AI in daily life requires setting boundaries and determining what tasks are worth automating and what should be kept under human control, and this is a personal decision that each individual must make for themselves 42s.
• One important consideration is the potential for AI to generate low-quality or misleading content, such as a poorly written book, which can damage a person's brand and reputation, and this is something that individuals should be aware of when using AI 42s.
• The line between what should be automated and what should be kept under human control is a personal one, and individuals must decide for themselves what tasks they are comfortable outsourcing to AI, such as using AI to assist with writing but not to replace human creativity and personality 42s.
• The individual has drawn a line on how much of their life they are willing to let AI take over, preferring to maintain control over creative tasks such as writing and structuring newsletters, and enjoying the struggle of putting scripts together 10s.

AI's Impact on Family and Personal Life

• AI is seen as useful for administrative tasks such as reading through insurance plans and looking at contracts, which can be time-consuming and tedious for humans 2m6s.
• The individual's family, including their wife and kids, has been affected by the experiment of letting AI take over their life, with the kids sometimes acting like robots and questioning the role of AI in their lives 4m42s.
• The individual's kids have been exposed to various AI technologies and are now more aware of AI and its potential impact on their lives, with even a four-year-old talking about AI and its potential applications 6m15s.
• A key takeaway from the experiment is that AI makes mistakes and it is essential to be willing to question the results and not blindly rely on AI, highlighting the importance of human oversight and critical thinking 10m42s.

Reflections and Lessons from the AI Experiment

• Another important point is that AI is not as smart and interesting as humans, and while it can be useful in certain contexts, it is essential to maintain a critical perspective and not overestimate its capabilities 12m10s.
• It is suggested that humanity should be relied upon more than other alternatives, as it is believed to provide better answers in the end 0s.

Closing Remarks and Credits

• A book is mentioned, with the author receiving praise for doing a great job, and gratitude is expressed to the author 10s.
• The interviewer is also commended for their work, although it is stated that a different moderator may be chosen for future events 42s.
• The production team for the show includes Christian Castro, Michelle Aloy, Katherine Milsop, Megan Bernie, and Kaitlin Lynch, with Nishot Kerwa serving as Vox Media's executive producer of podcasts 1m30s.
• Special thanks are given to Rosemary Hoe and Julia Sharp Lavine, and the theme music is credited to Trackademics 2m6s.