Artificial General Intelligence
Boston University - Spring 2025
Instructor: Prof. Iddo Drori, CDS 839
Pre-requisite: ML or Deep Learning or AI or Computer Vision or NLP
Grading: Presentations 50%, competition 40%, participation 10%
Martin Luther King Jr. Day (Monday, January 20): Holiday
Lecture 1 (Wednesday, January 22): Reasoning foundation models
Reading:
Chain-of-thought prompting elicits reasoning in large language models, Wei et al, 2022
Quiet-STaR: Language models can teach themselves to think before speaking, Zelikman et al, 2024
Rewarding progress: Scaling automated process verifiers for LLM reasoning, Setlur et al, 2024
Deliberative alignment: Reasoning enables safer language models, Guan et al, 2025
Towards system 2 reasoning in LLMs: Learning how to think with meta-chain-of-thought, Xiang et al, 2025
AlphaLLM: Toward self-improvement of LLMs via imagination, searching, and criticizing, Tian et al, 2024
rStar-Math: Small LLMs can master Math reasoning with self-evolved deep thinking, Guan et al, 2025
Free process rewards without process labels, Yuan et al, 2024
DeepSeekMath: Pushing the limits of mathematical reasoning in open language models, Shao et al, 2024
Process reinforcement through implicit rewards, Cui et al, 2025
OpenAI o3, Chen, Ren, Altman, 2024
DeepSeek-R1: Incentivizing reasoning capability in LLMs via reinforcement learning, DeepSeek-AI, 2025
Lecture 2 (Monday, January 27): Self-improving foundation models
Reading:
Self-improving foundation models without human supervision, Setlur et al, 2024, see References
Lecture 3 (Wednesday, January 29): Open source and open weights foundation models
2 OLMo 2 Furious, Walsh et al, 2024
DeepSeek-V3 technical report, DeepSeek-AI, 2024
Sky-T1, NovaSky, 2025
OLMoE: Open Mixture-of-Experts Language Models
Lecture 4 (Monday, February 3): World and self models
Reading:
Genie 2: A large-scale foundation world model, Parker-Holder et al, 2024
Lecture 5 (Wednesday, February 5): AI agents
Reading:
Building effective agents, Anthropic, 2024
Open source tools for building AI agents:
Model context protocol (MCP), Anthropic, 2024
Rivet: Visual workflows
LangGraph: Graph workflows
CrewAI: Multi-agent systems
OpenAI Swarm: Multi-agent orchestration
Trading agents: Multi-agents LLM financial trading framework, Xiao et al, 2024
Generative agent simulations of 1,000 people, Park et al, 2024
Lecture 6 (Monday, February 10): Generation, synthetic data, and verification
Multiagent finetuning: Self improvement with diverse reasoning chains, Subramaniam et al, 2025
Lecture 7 (Wednesday, February 12): Invited talk: Auto-continual learning, Prof. Joaquin Vanschoren
President's Day (Monday, February 17): No classes
Lecture 7 (Tuesday, February 18): Weight-space learning
Lecture 8 (Monday, February 24): Small, adaptive, fast learning, self-training foundation models
Reading:
Improving pretraining data using perplexity correlations, Thrush et al, 2025
LLM monorepo loss
Lecture 9 (Wednesday, February 26): AI for super-human Mathematics
IMO, autoformalization, Lean, test-time reinforcement learning
Reading:
Goedel-Prover: A frontier model for open-source automated
Theorem proving, Lin et al, 2025
STP: Self-play LLM Theorem provers with iterative conjecturing and proving, Dong and Ma, 2025
Gold-medalist performance in solving Olympiad Geometry with AlphaGeometry2, Chervonyi et al, 2025
Lecture 10 (Monday, March 3): AI for super-human visual reasoning
ARC, meta-learning, synthetic data, program synthesis, test-time training, human video screen analysis
Lecture 11 (Wednesday, March 5): AI for scientific discovery
Reading:
Can LLMs generate novel research ideas?, Si et al, 2024
The AI scientist: Toward fully automated open-ended scientific discovery, Lu et al, 2024
PaperQA2: Language agents achieve superhuman synthesis of scientific knowledge, Skarlinski et al, 2024
AutoSurvey: Large language models can automatically write surveys, Wang et al, 2024
Review feedback agent: Assisting ICLR 2025 reviewers with feedback, Thakkar et al, 2024
The virtual lab: AI agents design new SARS-CoV-2 nanobodies with experimental validation, Swanson et al, 2024
Spring recess (Saturday, March 8 - Sunday, March 16)
Lecture 12 (Monday, March 17): Video and audio analysis and synthesis, simulation
Lecture 13 (Wednesday, March 19): Meta learning, AI coding AI
Lecture 14 (Monday, March 24): Humanoid robots
Robots that learn, Malik, 2024
Isaac GR00T, Nvidia, 2024
Lecture 15 (Wednesday, March 26): Fly brain
Reading:
Neuronal wiring diagram of an adult brain, Dorkenwald et al, 2024
The fly connectome reveals a path to the effectome, Pospisil et al, 2024
Predicting visual function by interpreting a neuronal wiring diagram, Seung, 2024
Lecture 16 (Monday, March 31): Human brain
Human brain development, forebrain, midbrain, hindbrain, spinal cord.
Cerebral cortex: frontal lobe, prefrontal cortex, parietal lobe, temporal lobe, Boca's area, auditory cortex, occipital lobe, Wernicke's area.
Brain cells: Neuron, dendrite, microglia, oligodendrocyte, astrocyte, axon, synapse.
Blood-brain barrier: Astrocyte, transporter, endothelial cell, pericyte.
Pituitary gland: Hypothalmus, pituitary stalk, posterior lobe, capillaries, anterior lobe, thyroid.
Surrounding layers: Arachnoid mater, dura matter, blood vessels in subarachnoid space, pia mater, connective tissue, white and gray matter.
Self-cleaning: Cerebrospinal fluid, astrocyte, waste removal.
Neurotransmitters: Adrenaline, noradrenaline, dopamine, oxytocin, GABA, acetylcholine, glutamate, endorphins, serotonin.
Emotions: Anterior and posterior cingulate cortex, parahippocampal gyrus, hypothalmus, hippocampus, amigdala, septal nuclei.
Placibo: Cingulate cortex, insula, thalamus, nucleas accumbens, amygdala, ventral tegmental area, periaqueductal gray matter.
Memory: Frontal lobe, putamen, amygdala, temporal lobe, hippocampus, cerebellum, parietal lobe, caudate nucleus, thalamus.
Memory types: Sensory, short-term, long-term procedural, declarative, episodic.
Eyes: Optic nerve, optic chiasma, optic tract, thalamus, membrane, occipital lobes.
Music: Parietal lobe, frontal lobe, superior temproal gyrus, planum temporale, auditopry cortex, limbic circuit, insula.
Timing: Cerebellum, basal ganglia.
Consciosness
Ethics
Reading:
The unbearable slowness of being: Why do we live at 10 bits/s?, Zheng and Meister, 2024
Simulation and assimilation of the digital human brain, Lu et al, 2024
Lecture 17 (Wednesday, April 2): Human-AI co-evolution
Lecture 18 (Monday, April 7): Self-improving AI researcher
Reading:
A self-improving coding agent, Robeyns et al, 2025
PaperBench: Evaluating AI’s ability to replicate AI research, Starace et al, 2025
Lecture 19 (Wednesday, April 9)
Lecture 20 (Monday, April 14)
Lecture 21 (Wednesday, April 16)
Lecture 22 (Wednesday, April 23): Social responsibility
Lecture 23 (Monday, April 28): Safety and security
Reading:
AI Index report, Stanford, 2025
AI 2027 site, manuscript
Responsible scaling policy, Anthropic, 2024
OpenAI model spec, 2024
Scaling automatic neuron description, Choi et al, 2024
How to evaluate control measures for LLM agents? A trajectory from today to superintelligence, Korbak et al, 2025
What succeeding at AI safety will involve, Sam Bowman, 2024
An approach to technical AGI safety and security, Shah et al, 2025
Lecture 24 (Wednesday, April 30): AI governance
Reading:
Hardware-enabled governance mechanisms, Kulp et al, 2024
Open problems in technical AI governance, Reuel et al, 2024