Introducing the New Colleague - ChatGPT
summary
Claims that GPT‑4 1.76 trillion parameters drives rapid LLM progress, shows token limits (3.5: 4k, GPT‑4: 8k) and that plugins/Enterprise stop using data for training. Demonstrations use Microsoft image creator, Copilot, greto, Testcraft and Functionize; automation code generated for a real project fails because internal abstractions and IDs aren’t known. Surprising concrete wins: generative assistance cuts a 1,300‑case test plan to reduced 1,300 test cases to ~400, fine‑tuning cost orders of magnitude less than full training.
AI-written summary · about
chapters (36)
- 0:00Introduction
- 1:52AI hype cycle and Gartner’s model
- 2:57AI history and Transformers
- 4:21GPT models and parameters
- 6:07How LLMs work
- 8:43Training vs inference
- 10:08Temperature and creativity
- 11:35Language model limits
- 12:36Hallucinations and fine-tuning
- 14:06Fine-tuning vs full training
- 14:28Data limits and updates
- 15:35Token limits and memory
- 16:20Prompt engineering
- 17:24Custom ChatGPT instructions
- 18:07Risks and ethics
- 19:34QA engineers and AI
- 20:58Free vs paid ChatGPT
- 21:19Generating test data
- 22:28Automated test cases
- 23:14Code review and debugging
- 24:19Test strategy by domain
- 25:03AI report generation
- 26:31AI emails and comms
- 27:17Automation code challenges
- 28:41AI tools for QA
- 29:23Data anonymization with Gretto
- 29:44Self-healing tests with Functionize
- 30:26Self-healing Selenium tests
- 30:47Test triggers and exploration
- 31:31AI test ideas with TestCraft
- 32:35Future of AI and automation
- 33:45AI’s job impact
- 34:46AI augmentation vs replacement
- 35:49Tech skill changes
- 36:54Adapting to AI
- 37:15Closing remarks
related talks
02h 00mChatGPT & Prompt Engineering
39mHow ChatGPT helps to boost your development efficiency
15mHarnessing the Power of Generative AI for Intelligent Applications
01h 09mChatGPT and OpenAI API solutions: 👑 My Favourite Implementations
48mTestGPT: Generative AI (AutoGPT, ChatGPT & VisualGPT)
01h 16mChatGPT - and How we Can Benefit from it
27mChatGPT Ignites
45mHow to use GPT in your projects: A Case Study