book

Introduction to Generative AI

by Numa Dhamani, Maggie Engler

February 2024

Intermediate to advanced

336 pages

9h 8m

English

Manning Publications

Audiobook available

Read now

Unlock full access

forewordprefaceacknowledgmentsabout this bookWho should read this bookHow this book is organized: A road mapliveBook discussion forumsOther online resourcesabout the authorabout the cover illustration
Evolution of natural language processingThe birth of LLMs: Attention is all you needExplosion of LLMsWhat are LLMs used for?Language modelingQuestion answeringCodingContent generationLogical reasoningOther natural language tasksWhere do LLMs fall short?Training data and biasLimitations in controlling machine outputsSustainability of LLMsRevolutionizing dialogue: Conversational LLMsOpenAI’s ChatGPTGoogle’s Bard/LaMDAMicrosoft’s Bing AIMeta’s LLaMa/Stanford’s AlpacaSummary
How are LLMs trained?Exploring open web data collectionDemystifying autoregression and bidirectional token predictionFine-tuning LLMsThe unexpected: Emergent properties of LLMsQuick study: Learning with few examplesIs emergence an illusion?What’s in the training data?Encoding biasSensitive informationSummary
Safety-focused improvements for LLM generationsPost-processing detection algorithmsContent filtering or conditional pre-trainingReinforcement learning from human feedbackReinforcement learning from AI feedbackNavigating user privacy and commercial risksInadvertent data leakageBest practices when interacting with chatbotsUnderstanding the rules of the road: Data policies and regulationsInternational standards and data protection lawsAre chatbots compliant with GDPR?Privacy regulations in academiaCorporate policiesSummary
The rise of synthetic mediaPopular techniques for creating synthetic mediaThe good and the bad of synthetic mediaAI or genuine: Detecting synthetic mediaGenerative AI: Transforming creative workflowsMarketing applicationsArtwork creationIntellectual property in the LLM eraCopyright law and fair useOpen source and licensesSummary

Cybersecurity and social engineeringInformation disorder: Adversarial narrativesPolitical bias and electioneeringWhy do LLMs hallucinate?Misuse of LLMs in the professional worldSummary
Using LLMs in the professional spaceLLMs assisting doctors with administrative tasksLLMs for legal research, discovery, and documentationLLMs augmenting financial investing and bank customer serviceLLMs as collaborators in creativityLLMs as a programming sidekickLLMs in daily lifeGenerative AI’s footprint on educationDetecting AI-generated textHow LLMs affect jobs and the economySummary
Chatbots for social interactionWhy humans are turning to chatbots for relationshipThe loneliness epidemicEmotional attachment theory and chatbotsThe good and bad of human-chatbot relationshipsCharting a path for beneficial chatbot interactionSummary
Where are LLM developments headed?Language: The universal interfaceLLM agents unlock new possibilitiesThe personalization waveSocial and technical risks of LLMsData inputs and outputsData privacyAdversarial attacksMisuseHow society is affectedUsing LLMs responsibly: Best practicesCurating datasets and standardizing documentationProtecting data privacyExplainability, transparency, and biasModel training strategies for safetyEnhanced detectionBoundaries for user engagement and metricsHumans in the loopAI regulations: An ethics perspectiveNorth America overviewEU overviewChina overviewCorporate self-governanceToward an AI governance frameworkSummary
The quest for artificial general intelligenceAI sentience and consciousness?How LLMs affect the environmentThe game changer: Open source communitySummary
Chapter 1Chapter 2Chapter 3Chapter 4Chapter 5Chapter 6Chapter 7Chapter 8Chapter 9