“Do I have something in my teeth?”
If AI were a first date, that’s how it would start—slightly awkward, wildly promising, and full of surprises. Welcome to Speed Dating with AI: each week I’ll “meet” a different AI, see what it’s great at, where it falls short, and whether it deserves a second date.
What is AI?
Artificial Intelligence is not magic. It’s data + algorithms + repeated learning cycles to perform tasks that usually require human intelligence.
Training vs. Inference (a two-phase relationship)
Training: the “gym phase.” The model learns from large datasets (costly, compute-heavy).
Inference: the “real-world phase.” The trained model answers, predicts, generates—fast and (ideally) efficient.
The Different Faces of AI
AI isn’t one thing; it’s a family of approaches:
Machine Learning (ML): learns patterns to predict (fraud, churn, recommendations).
Natural Language Processing (NLP): understands/generates language (chatbots, summarization).
Computer Vision: interprets images/video (quality control, medical imaging).
Robotics & Autonomous Systems: acts in the physical world (warehouse robots, self-driving).
Generative AI: creates new text, images, audio, video, or code.
What’s striking is that many of the AIs we already use daily—sometimes without even realizing—fall under these categories. From recommendation engines in streaming platforms, to voice assistants on our phones, to fraud detection running silently in the background, AI is already woven into our everyday lives.
From First Wave to Second Wave
Wave 1: LLMs went mainstream (ChatGPT, Copilot) and everyone “dipped a toe.”
Wave 2 (now): targeted adoption—teams pick specific use cases tied to data readiness, compliance, and ROI. Less hype, more fit-for-purpose.
And remember: everyone wants to date a surfer. If you missed the first wave, educate yourself and get ready for the second — because this one is about real, sustainable adoption.
What We’ll Explore (season menu)
This series will feature hands-on dates with tools that are shaping how we work and create:
ChatGPT – conversational AI that opened the door to mass adoption.
Google Cloud + Vertex AI – enterprise-grade tools for building, training, and deploying ML models.
Perplexity – a new approach to search and knowledge discovery through conversational interfaces.
NotebookLM – an experimental tool for turning documents into interactive, AI-powered assistants.
Freepik AI – creative generation for design and visual storytelling.
MidJourney – one of the most popular platforms for generative imagery, blending art and AI.
ElevenLabs – state-of-the-art voice AI for lifelike text-to-speech and custom voice cloning.
Runway ML – leading video generation platform, pushing the frontier of storytelling through motion.
Each week: what it does well, what it doesn’t, ideal use cases, data needs, risks, and a quick verdict (Second date? Friend zone? Hard pass?).
The Future: Bubble or Backbone?
Some predict the AI bubble will deflate—training costs, maintenance, and unclear ROI will force a correction. Others point to a reinforcing loop: AI drives demand for better chips, tooling, and efficiency (think NVIDIA & co.), which then makes AI more sustainable and useful.
My take: we’re in AI’s adolescence—brilliant, moody, and boundary-testing. That’s exactly why governance, measurement, and ethics matter now.