BACK

I. Introduction to AI Applications and Impact
- Over 100,000 people die annually from snake venom; AI-designed antivenoms can be cheaply produced in silico.
- AI contributes beyond medicine: material science innovation, decoding animal languages, and astrophysics (e.g., black hole imaging).
- AI helps uncover hidden historical texts (Vesuvius challenge) and animal communication (Earth Species Project).

II. Origin and Evolution of AI
- AI concept began in 1955 with efforts to enable machines to think like humans.
- Modern AI advances rely on massive data to improve performance in image and text recognition.
- Large language models predict words based on extensive internet text.
- AI now processes various data types: audio, video, medical images, 3D signals.
- Some AI models surpass human capabilities in specific tasks.

III. Current Research & Challenges
- Agentic AI can outperform humans in persuasion and economic incentives.
- Current AI benchmarks are largely inadequate; a new "humanity last exam" is assessing models with 2,500 scientific questions—top models still score low (~20% correct).
- Cultural biases exist in AI language models, which mostly reflect Anglo-Saxon linguistic patterns—need for multilingual and multicultural model development.

IV. Personal Experience & UN Work
- Speaker rooted in AI and neural networks, with experience at NASA and the UN.
- Worked on humanitarian AI applications: emergency mapping via satellite images, augmenting human analysts with AI for precision.
- Led COVID digital twin simulation for a 700,000-person refugee camp to test emergency strategies.

V. Societal and Ethical Considerations
- AI presents both opportunities and risks akin to nuclear technology.
- Public perception about AI is split (~50% excited, 50% nervous).
- Important considerations include AI’s carbon footprint and whether AI models should be centralized in the cloud or distributed locally.

VI. Future of AI and Generative Engineering
- Shift from deterministic mathematical models to probabilistic AI models.
- Future trends include generative engineering to design new machines and tools.
- Human-created content may gain new value amidst widespread AI-generated material.

VII. AI in Employment and Collaboration
- Job automation expected to impact cognition-based work; augmentation likely for 15-30% of jobs.
- Humans collaborating with AI ("centaurs" or "cyborgs") show improved performance.
- Studies indicate workers combined with AI outperform teams without AI.
- Key future skills include formulating the right questions rather than providing all answers.

VIII. Educational Transformation
- AI offers new learning methods, such as the Socratic approach via AI questioning.
- Encouragement to engage with AI for fast, deep learning in various topics.

IX. AI for Medicine: SpotLab Example
- AI used to analyze bone marrow images to diagnose blood cancers, improving precision beyond human capability.
- Combines patient DNA data and medical images to tailor treatments.
- Developed AI "filters" that work with smartphone microscopes for accessibility in low-resource settings (e.g., Ethiopia, Colombia).
- Interdisciplinary teams of medical doctors and AI experts are vital for effective AI applications.

X. Broader Reflections and Call to Action
- Future projects should be anti-disciplinary or AI-disciplinary, blending ideas across fields.
- AI will become foundational infrastructure globally; need to preserve cultural diversity.
- Leaders and individuals must act as "orchestra directors" of algorithms, ensuring AI serves human values.
- Emphasis on using AI to enhance humanity and address meaningful challenges, especially for future generations.

Actionable Items and Tasks:
- Develop culturally and linguistically diverse AI foundational models.
- Design robust AI benchmarks beyond current standards like the "humanity last exam."
- Integrate human-AI collaboration frameworks to augment jobs, especially cognitive work.
- Promote interdisciplinary teams combining domain experts and AI specialists.
- Deploy AI medical diagnostic tools accessible in resource-limited regions using smartphone-based solutions.
- Incorporate AI literacy and engagement in education systems embracing new learning paradigms.
- Assess and optimize AI models for environmental impact and energy consumption.
- Foster ethical AI deployment prioritizing positive societal impact and human augmentation.

AI for Humanity Grand Challenges

Share:

13:50 - 14:10, 27th of May (Tuesday) 2025 / INSPIRE STAGE

AI is reshaping our economies and society—but the real challenge is using it to serve humanity and the planet. From universal health coverage to response to disasters, this talk explores how AI can help solve global challenges, if we ask the right questions and build with purpose. We will also explore potential future scenarios of AI, its impact in jobs, and how to shape AI direction guided by a compass rooted in human values.

AUDIENCE:
Profitable Company
TRACK:
AI/ML Business & Tech trends Tech4Good