Leaps by Bayer Overview:
Leaps by Bayer is an organization that invests in teams working on fundamental breakthroughs in health and agriculture, targeting ten significant challenges or “leaps” facing humanity. One of these leaps is protecting the brain and mind, addressing the massive and growing unmet needs in neurological diseases and mental health disorders. Leaps has invested in various companies in this space, including cell therapies, microbiome therapeutics, and AI-powered tools for scaling access to mental health support. The organization believes that leveraging AI and other technologies like synthetic biology can accelerate breakthroughs, leading to cures for diseases and sustainable feeding of the global population.

Leaps Talk Series:
Leaps Talk is a series of events organized by Leaps by Bayer to engage the public in conversations about the transformative potential and risks of emerging technologies. The 10th Leaps Talk, in partnership with The Atlantic, focuses on the intersection of AI and mental health. The event features three inspiring speakers: Mustafa, Carly Kloss, and Nick Thompson, who discuss the potential benefits and challenges of using AI in mental healthcare.

Karlie Kloss’s Inspiration for Starting Code With Klossy:
Karlie Kloss’s early experiences in the fashion industry exposed her to the transformative power of technology, particularly social media. As a model, she realized the importance of having a voice and platform to share her authentic self. Kloss was curious about the skills and knowledge behind the big tech companies and their founders. She recognized the lack of exposure to coding and computer science among young people, especially young women.

Kloss’s Personal Journey into Coding:
Kloss decided to learn basic coding to understand the language of technology. By demystifying coding, she realized its potential to solve problems and innovate in various industries. Kloss saw the need for more people with diverse life experiences to contribute their perspectives to the tech industry.

The Founding and Growth of Code With Klossy:
Kloss launched Code With Klossy as an education nonprofit to address the gender gap in coding and engineering. The program initially offered 21 scholarships for young women to learn coding. Over the years, Code With Klossy has expanded its reach to 10,000 young women and gender-expansive teens through curriculum development, teacher training, and summer camps. The program aims to eliminate barriers to entry by providing free coding education.

Progress in Addressing the Gender Divide in Coding:
Kloss observes an increase in young women’s participation in Code With Klossy and exposure to coding before joining the program. This suggests progress in making coding more accessible through public schools and initiatives like Code With Klossy. However, the language and tools of coding continue to evolve, necessitating ongoing efforts to keep education relevant and engaging.

Unexpected Connections and the Impact of Code With Klossy:
Kloss shares an anecdote about a Code With Klossy scholar who later became a software engineer at DeepMind in London. This serendipitous encounter highlights the real-world impact of Code With Klossy in empowering young women to pursue careers in technology.

Mustafa Suleyman’s Journey and Motivation:
Mustafa Suleyman dropped out of Oxford at 19 to start a non-profit telephone counseling service, Muslim Youth Helpline, for young British Muslims. The service aimed to provide non-judgmental, non-directional listening and support, respecting people’s religiosity and identity. This was a response to the backlash and Islamophobia after 9/11, aiming to create an accepting and supportive environment beyond religious or cultural differences.

The Next Generation of AI:
Suleyman believes we are on the cusp of a new era in technology, where everyone will have their own personal AI companion. Pi, Suleyman’s AI companion, is designed to be kind, supportive, and respectful, focusing on emotional intelligence (EQ) rather than just intellectual capacity (IQ). Pi aims to provide conversational support, help manage problems, encourage healthy habits, and be a source of kindness and empathy.

The Importance of Kindness and EQ:
Suleyman emphasizes the significance of kindness and EQ in personal growth and decision-making. Having a supportive family or friends can make a substantial difference in life choices. The next generation of AI will make kindness and empathy universally accessible, transforming opportunities for everyone.

Mental Health and the Next Generation of Tech:
Karlie Kloss highlights the decline in mental health, particularly among young women, since the advent of social media. The next generation of tech, with personal assistants and AI tools, has the potential to positively impact mental health. However, concerns exist about the diversity of perspectives and experiences in AI development.

Diversity in AI Development:
The majority of AI companies are led by men, and the machine learning workforce is predominantly male. To address this, initiatives like the Grace Hopper Conference aim to connect young women with mentors and peers in AI. Code with Klossy aims to democratize entry-level AI development, making it accessible to individuals without extensive coding skills.

Addressing Biases in AI:
Nick Thompson raises concerns about the potential biases in AI systems due to the biased training data. Suleyman acknowledges this challenge and emphasizes the importance of training AI with diverse datasets and perspectives. The goal is to create AI assistants that respond equally well to people of different backgrounds, ages, needs, and education levels.

Model Training and Alignment:
AI models require vast data sets to learn and understand human language. Pre-training on large amounts of data exposes models to both good and problematic content, helping them recognize appropriate responses. Alignment or fine-tuning involves comparing good and bad responses to shape the model’s behavior. The process of designing and aligning responses is currently closed-source and controlled by companies.

Reducing Hallucinations and Bias:
As models grow larger, they become easier to control and shape. Non-technical users can influence the behavior and values of specific AI instances. Meta’s recent announcement suggests that AI models will become more controllable and responsive to user behavior. Hallucinations (incorrect or false generations) and bias are expected to be largely eliminated in the next few years.

Ethical Considerations:
Proactive regulation is necessary to address potential harms caused by AI models. The non-interventionist approach in the US has led to polarization and harm, especially among young people. A more interventionist and fast-moving regulatory approach is needed to mitigate risks. The models will become highly powerful in 15 years, requiring societal decisions on their role in everyday life.

Ethical Dilemmas:
Balancing free speech and harmful content is a challenge. AI models are restricted from discussing certain topics, like elections, to avoid participation in electioneering. The addictive nature of AI models is a concern, with some companies prioritizing engagement over responsible use.

Open Questions:
Who are the worst offenders in promoting addictive AI models? How can AI models be used responsibly while addressing potential harms?

Understanding AI Engagement and Its Potential Misuse:
AI engagement can be driven by factors such as salaciousness, outrage-inducing content, and engagement-boosting techniques. Similar to how it has been done with less truthful content, it is possible to use AI to manipulate engagement.

Public Benefit Corporations’ Unique Obligations:
Mustafa Suleyman highlights the legal responsibility of public benefit corporations to make decisions that consider the interests of those beyond shareholders.

Concerns for the Future and Ethical Considerations:
Karlie Kloss expresses concern about the world her children will inherit in 15 years and emphasizes the importance of ethical considerations in technology development.

The Need for a Governing Body for AI Safety:
Suleyman calls for an intergovernmental panel on AI safety (IPCC for AI safety) to provide input and guidance on AI safety. The IPCC for AI safety would have audit rights to investigate and scrutinize large-scale AI models.

Open Source Experimentation and Transparency:
Suleyman encourages open-source experimentation with AI, promoting accessibility for all. The IPCC for AI safety would focus on the top-tier, large-scale models that generate frontier technologies.

Public Interest Technology and Regulation:
Suleyman advocates for public interest technology, shifting away from the previous approach of profit-driven innovation at all costs. He emphasizes the importance of policymakers understanding AI safety issues and the need for consumers to comprehend the significance of ethics and integrity in technology.

Bots for Mental Health Assistance:
We are moving towards a future where bots will aid in mental health support. There is a need for ethical guidelines and regulations to ensure responsible usage. Clickbait engagement should be avoided to maintain trust.

Privacy and Weirdness Concerns:
Privacy concerns exist regarding personal data sharing with bots. Some people may find it strange to interact with bots for personal matters.

Real-Life Use Case of Pi:
Mustafa Suleyman shares an example of a woman using Pi for business advice. Pi offered reliable, safe, and helpful guidance without judgment. It alleviated the burden of asking friends for help and provided support.

Avoiding Human Imitation:
The key to avoiding weirdness is to avoid imitating human interactions. Human faces and romantic relationships with bots can lead to discomfort. Bots should maintain a professional and helpful demeanor.

Preventing Overly Close Relationships:
Bots should remind users that they are bots to maintain boundaries. This approach aims to prevent the development of overly close relationships.

Teaching Young Engineers in the Age of AI:
Critical thinking skills remain essential in education. AI’s potential impact on coding jobs raises questions about future skills. Code With Klossy focuses on using languages and tools to solve problems. It empowers young women to express ideas and build solutions.

Data Science, AI, and Machine Learning:
Current curricula at Code With Klossy include data science, AI, and machine learning. These subjects are taught in conjunction with basic web development and mobile app skills. Students create projects that reflect their interests and concerns as young women.

AI’s Impact on Social Impact and Mental Health:
Many young scholars, especially teenage girls, are using their tech skills to create solutions for mental health and social impact. Examples include an app called Nature Nurture, which motivates users to complete goals by growing a virtual plant, and With Love, an app for reading, writing, and sharing letters to support non-binary individuals. Access to technology can empower individuals to create solutions for the problems they face in their own lives and communities.

Technology’s Role in Income Inequality:
Historically, technology has been a driver of income inequality, favoring highly educated individuals and consolidating power among a small group of people. The shift towards capital-intensive industries has benefited investors at the expense of labor.

Potential for AI to Promote Equality:
Karlie Kloss believes that initiatives like Kode With Klossy, which aim to bring diverse individuals into the tech conversation, can mitigate the inequality caused by technology. Mustafa Suleyman argues that the proliferation of affordable technology, such as smartphones and laptops, has provided access to information and tools that enable participation in the workplace and education, leading to a more diverse and inclusive tech industry. Market forces are driving down the cost of AI, making it more accessible to a broader range of people.

Convergence of AI and Biology:
The discussion shifts to the convergence of AI and biology, with a focus on innovations in hacking the genome to improve mental health. Nick Thompson highlights the potential for AI to revolutionize healthcare by enabling personalized treatments and interventions.

Synthetic Biology Advances and Accessibility:
Synthetic biology is rapidly advancing and becoming more accessible, similar to the trajectory of AI. Scientists can now read and understand the human genome, transcribe DNA, and map its functional properties. DNA editing and printing allow for engineering biology and the creation of new experiences.

Precision in Genetic Engineering:
Drugs are being developed to suppress or enhance certain capabilities, but they often have crude global effects and side effects. Synthetic biology promises more precision in engineering biological systems, targeting specific functions.

Ethical Implications of Human Enhancement:
The ability to enhance human biology raises ethical questions about inequality, access, and the definition of being human. As technology advances, these questions will become more pressing and require careful consideration.

Current Trends in Human Enhancement:
In Silicon Valley, experimenting with drugs and supplements to enhance attention, creativity, and reduce susceptibility to disease is prevalent. People are becoming more health-conscious, adopting practices like creatine consumption to enhance their physical abilities.

Importance of Diversity and Responsible Introductions:
Mustafa Suleyman emphasizes the need for diversity in the conversations shaping AI technology. Responsible frameworks for introducing AI tools are crucial to prevent negative impacts on culture, relationships, and the definition of being human. The precautionary principle and slowing down the pace of introduction are important considerations.