Target audience: Senior leadership / C-suite level
Maximum participants: 20
3-hour version (one day)
- About Greg Baker, his research, expertise, and areas of ignorance
- AI's current and future impact on jobs (5-year horizon)
- AI in 2029: Predictions and projections
- Strategic decision-making for technology (Wardley mapping and two types of strategy)
- Evaluating and working with AI vendors: Warning signs of under-delivery and over-promising
- Choice of: "How to evaluate experiments in-house quickly" or "How to see the future and quantify it"
6-hour version topics (two days):
All of the 3-hour version topics, plus:
- Baumol's cost disease and baked-in 25% salary rises for some industries
- Implications of the prompt-engineering vs fine-tuning debate ($20 trillion at stake)
- Choice of: "How to prompt engineer" or "Alternative principles of AI ethics"
- Exploring the AI zoo: Strengths of ChatGPT, Microsoft Copilot, Anthropic, Gemini, Llama and Mixtral
- AI's limitations and shortcomings
- Recent advancements in speech, listening, and translation (last 18 months)
- Agentic AI, asynchronous bots, custom bots, tool calling, retrieval augmented generation, synchronous interaction (demos, vendor discussions, resources)
- Governance frameworks, capabilties and processes needed for decentralised AI adoption
- Identifying opportunities in your organization (impromptu deep dive on implementation, vendors, pitfalls, etc.)
- Implications for management
For questions or inquiries, please email [email protected].
Organisational AI Enablement
Introduction to Generative AI
Target audience: Team leads and individual contributors
Topics covered (1 hour)
- Where GenAI helps
- Problems
- Capabilities: create, summarise, inspire, critique, educate, role-play, automate, translate, analyse, watch
- Speech and vision
- Managing and empowering staff to use AI
Prompt Engineering
Target audience: Team leads and individual contributors
Topics covered (1 hour)
- How GenAI systems are trained
- An intuition on how prompting works
- What is temperature, and what does it do
- Giving instructions versus assuming a role
- 7 different levels of expertise
- Problems: arithmetic, hallucination, thinking ahead, prompt injection
- Markdown, templates, registers and other magic tricks that do work
- Alignment, the refresh cycle, and why most other tricks don't work
- Getting around refusal
- How to get good at prompting
Ethics of AI
Target audience: Staff in governments and non-profits concerned with AI's impact on diverse communities.
Topics covered (1 hour)
- Accountability and responsibility
- Transparency and explainability
- Data privacy
- RLHF and where GenAI gets its moral sense from
- Prompting for fairness reading
- Environmental impact
- Hallucinations and explanations
- Issues when working in other languages
- Ethics of AI from different religions and perspectives
These courses are mostly tuned for users of OpenAI ChatGPT
and Microsoft Copilot. Versions for Gemini, Anthropic Claude,
and local-run models are in development.
For questions or inquiries, please email [email protected].
Software Developer Upskilling in AI
Minimum Viable Software Developer for Text and Language
Target audience: Software developers (any level)
Topics covered (20 hours):
- Calling ChatGPT and Anthropic APIs
- Tool calling
- Retrieval augmented generation (RAG)
- Building information retrieval systems
- Handling different languages and scripts
- Generating audio with ElevenLabs
- Integrating Twilio for telephone bots
On-prem, On-device, On-budget
Target audience: Software developers with experience in OpenAI/Anthropic/Gemini APIs
Topics covered (20 hours):
- Common tasks in text and voice
- Using and running pre-existing models
- Fine-tuning existing models
- Model compression techniques
- Design principles for language applications
For questions or inquiries, please email [email protected].