ChatGPT, Gemini, & LLM Masterclass - From an AI expert

Compare the three giants of generative AI, Gemini, ChatGPT, and Copilot, across facts, logic, maths, content generation, multimodality, and coding, while reviewing their paid plans and integrations.

Compare how Gemini, Copilot, and ChatGPT handle facts and research, including internet access, source citation, and information structure, with a verdict ranking Gemini first, Copilot second, and ChatGPT third.

Compare summarization and long inputs across ChatGPT and Gemini Advanced, examining token limits, quality, style, and control, with Copilot’s lower limit, using Oliver Twist.

Navigate ChatGPT's interface from sign-up and login to model options like GPT-3.5 and GPT-4, and learn to plan trips with ideal dates using concise, bullet-point prompts.

Explore Bard, Google's version of ChatGPT, its extensions with Google Flights, Maps, and Workspace, and compare its creativity, explainability, and factuality to GPT-3.5 and GPT-4.

Master prompt engineering fundamentals and the basic toolbox for large language models. Craft clear, specific prompts, break large tasks into steps, and prevent hallucinations.

Explore prompt patterns as reusable structures that shape interaction, style, behavior, and output by assigning roles, contexts, goals, instructions, output formats, examples, and input data to large language models.

Explore the role pattern in prompts by assigning a persona to a large language model, guiding it to act as an expert in neurology, copywriting, or digital marketing.

Learn how giving context and a clear goal shapes output, including role assignment, audience details, and objectives to inform, connect, and sell, with practical newsletter examples.

Focuses on the instruction portion of prompts, teaching clear, specific commands, avoid fluff, specify lengths, break tasks into simple steps to control large language model's behavior, transformation and summarization.

Apply the output control pattern to define how a large language model answers, shaping format, style, and audience with formats like bullet lists, tables, or JSON.

Explore zero-shot and few-shot prompting, in-context learning, and tuning to guide large language models with examples, instructions, and final inputs for effective summarization and distinguishing salesy, convincing, and dull language.

Separate input data with clear delimiters to bound external data, such as text excerpts or links, improving reliability and auditable provenance while enabling search-engine style, up-to-date responses.

Explore advanced prompt engineering by examining four prompt types: augmented interaction, optimization, reasoning, and long tasks, and how they enable interview prep, learning topics, and long-form writing projects.

Learn how root prompts shape stateful language models across an entire conversation, define answer style and detail, enforce policies, and reset instructions for building software with LLMs.

Use the question refinement pattern to have the language model propose a better version of each question before answering, optimizing for your objective and improving accuracy.

Explore the flipped interaction pattern where the model asks questions to guide you toward a tailored plan, illustrated with a London ride-sharing business plan and social twist.

Utilize a gameplay pattern where the language model acts as a game master, applying rules, asking questions, providing feedback, and enabling interactive learning of Spanish, data science, and interview preparation.

Learn to optimize prompts with the meta language pattern using symbolic notations to condense instructions for travel planning and scheduling, improving reuse and maintainability.

Explore the output automator pattern to generate automated outputs based on conditions, such as bilingual responses and language-specific code. Apply steps to create multi-step algorithm outputs in Python and Java.

Explore the alternative approaches pattern to generate multiple solutions, compare pros and cons, and expand problem-solving horizons while mitigating cognitive bias through example workouts and code strategies.

Explore how large language models rely on parameter counts to learn abstractions, reason, and plan. Discover distillation, step-by-step reasoning, and tuning with plugins and functions to boost performance.

Learn how to use a fact-check list prompt to surface underlying assumptions, verify key facts, and compare options with concise reasoning for AI answers.

Discover how the cognitive verifier prompt breaks a question into sub questions, gathers inputs, and recombines reasoning steps for a more accurate final answer.

Explore the reflection pattern that adds the reasoning and assumptions behind each answer, enabling validity checks and clearer explanations, with examples of GDP and knowledge cut-off date.

Master chain-of-thought prompting to boost large language model reasoning by breaking problems into steps. Explore 'let's think step by step' and one-shot to few-shot prompts to improve accuracy with examples.

Apply least to most prompting to guide the language model through iterative breakdown of problems, similar to chain of thought, enabling you to solve more complex problems.

Leverage self-consistency by running multiple chain-of-thought prompts for the same problem, sampling diverse reasoning paths, and selecting the most frequent final answer to improve accuracy on complex problems.

Explore tree of thoughts, building on chain of thought and self-consistency by sampling multiple reasoning paths to find the best global solution for complex problems, like the game of 24.

Master the react reasoning and action framework to enable language models to break problems into steps and actions using tools like search, read a pdf, and read a url page.

Understand token size limits in chat models, compare input and output tokens across GPT-4, GPT-3.5, Bard, and Claude, and apply techniques to manage long texts and large data corpora.

Outline a long task and iteratively expand each part with feedback to refine blogs, code, emails, or books.

Demonstrate tail generation to keep instructions visible by ending each answer with repeats of key instructions, and explore outlining, token limits, stateful models, and efficiency in long tasks.

Learn to process large inputs beyond token limits through iterative chunking and summarization. Compare sequential and parallel approaches, using overlap and natural chunks to maintain context.

Ground large language models in a trusted data corpus by building a searchable dataset. Connect data to the model with vector databases and embeddings to generate sourced answers.

Explore how Chrome extensions boost ChatGPT summarization across web pages using tools like summarize everything, with customizable prompts, emoji bullets, and batch summarization within token limits.

Explore advanced chrome extensions like cider, a ChatGPT sidebar that unlocks multiple language models, chat, compose and reply tools, OCR, PDF chat, and custom prompts to boost productivity.

Discover how ChatGPT plugins extend language models by granting tools like calculators, web access, Wikipedia, and PDF readers, and learn how prompts trigger these plugins.

Learn to enable and manage ChatGPT plugins, choosing between default gp4 and plugin versions. Explore plugins like Wolfram, Link reader, Wikipedia, and Expedia for fresh data and web access.

Explore token size limits in OpenAI models, covering input and output tokens, model comparisons, costs, and practical techniques to manage long texts, long outputs, and large data corpora.

Master the OpenAI API completion through a hands-on walkthrough, adjusting model, temperature, max length, top-p, stop sequences, and zero-shot, one-shot, and few-shot prompts and penalties to control output and creativity.

In the OpenAI chat walkthrough, define a chatbot’s persona with the system prompt, including audience persona and output style, and apply the basics of prompt engineering to manage context.

Explore how an ai assistant grounds answers with documents, retrieval, and functions, routes via per-run instructions, and cites sources for grounded stock and delivery queries.

Explore leveraging large language models like ChatGPT and Bard to scale content from emails to e-books, using the IOE framework to preserve originality, voice, and intellectual honesty.

In ideation phase, define the topic and decide what to say, drawing on your own ideas, LM-driven brainstorms, and insights from existing content with citations to yield a unique voice.

Create a coherent, source-of-truth outline for long content by defining the objective and audience, setting the structure and level of detail, and organizing input ideas into themes and sections.

Master the execution phase by defining the language model persona, objective, audience, and writing style, then iteratively expand your outline with feedback to produce final, polished content.

Explore methods to control writing style, from describing with adjectives to extracting from good examples and following best-practices prompts for short-form tweets and longer content.

Explore how ChatGPT and Gemini power Twitter and X content, blending viral and informative posts, crafting short punchy tweets, threads, and replies with targeted prompts.

Master authentic Instagram strategy with images, reels, stories, and messages to build trust and influence purchases, guided by ChatGPT and other LLM prompts for planning and launches.

Learn to craft Instagram prompts with clear roles, audience, and goals; build punchy, fact-based posts and five-paragraph threads, use hooks and emotions, and generate imagery with DALL-E.