Where is AI Today & Where is it Going
Class 1, Video 1: Where is AI Today and Where is it Going?
Welcome to AI Readiness for Go-To-Market Teams. In Class 1, we’re going to build a shared foundation so the rest of the certification stays practical, consistent, and easy to apply. We’ll cover three things across the class: where AI is today and where it’s going, what GTM teams can do in the age of Generative AI to adapt, and the RPX point of view and next steps.
Why this section matters
There’s a lot of noise online about AI. This section is meant to cut through that by grounding in high-authority sources and giving you an “enough to be dangerous” foundation without turning you into a machine learning engineer.
Start with the basics: intelligence, learning, and machine learning
To understand artificial intelligence, we first need simple definitions:
- Intelligence is the ability to acquire and apply knowledge.
- Learning is the process of acquiring knowledge, skills, or understanding through experience or study.
- Machine learning is a method that allows computers to acquire capabilities that look like learning, reasoning, and decision-making.
Machine learning does not work like human memory. It’s a different mechanism. For this certification, you don’t need to become technical about it, but it helps to know that AI behavior and human behavior are fundamentally different under the hood.
What is AI?
AI is an umbrella term. It’s a field of computer science focused on building systems that can learn, reason, make decisions, and perform tasks that normally require human intelligence.
You’ll see similar wording across major definitions: AI systems simulate learning, comprehension, problem solving, decision-making, and creativity. The key idea is consistent: AI aims to perform “intelligent” tasks at a machine level.
Types of AI (big picture)
A useful way to keep AI categories clean is to separate what exists today from what’s theoretical:
- Artificial Narrow Intelligence (ANI): This is the only real, usable AI we have today. It does specific tasks.
- Artificial General Intelligence (AGI): A theoretical AI that would perform across tasks at a human-level generality.
- Artificial Super Intelligence (ASI): A further theoretical stage beyond AGI.
For GTM leaders, the important takeaway is simple: everything you’re using today is narrow AI. Powerful, yes. But not “general intelligence.”
Types of AI by functionality (practical framing)
Another helpful lens is: how does the system behave?
- Reactive systems: Fixed-logic AI that responds to a specific situation and task (classic examples include older game bots and rigid decision systems).
- Limited learning systems: Models that detect patterns and make predictions/generations (this includes many predictive and generative systems).
- Contextual systems: AI that uses recent information to adapt decisions in near real time (examples include some virtual assistants and autonomy use cases).
This is not meant to be a perfect taxonomy. It’s meant to keep your thinking organized so you can place AI tools into a “behavior bucket” and predict risks.
A quick history of AI (why we’re here today)
A very high-level timeline:
- The Turing Test was an early milestone asking: can machines imitate human thinking convincingly?
- Deep Blue was a major moment where a machine beat a human chess champion.
- AlexNet was a key moment that accelerated modern AI progress (worth looking up for deeper history).
- AlphaGo showed major progress in complex strategy and learning.
- The Transformer architecture (2017) became a foundational breakthrough that later powered modern large language models and the GenAI boom.
This is why GenAI feels “sudden.” The earlier seeds were planted years before the public explosion.
Core learning methods (what you’ll hear referenced)
You’ll often hear “AI” and “machine learning” used together. In practical terms, machine learning is the approach used to create AI models. Four methods worth knowing:
- Supervised learning: Learns from labeled examples (spam filtering, credit scoring, basic recognition tasks).
- Unsupervised learning: Finds patterns in unlabeled data (clustering, recommendations, anomaly detection).
- Reinforcement learning: Learns through trial and error (often used in games, autonomy, and some advanced systems).
- Self-supervised learning: Trains from large unlabeled data where the model generates its own learning signals (this is relevant to modern LLMs like ChatGPT, Gemini, Claude, etc.).
Where AI is today (the reality as of now)
AI has moved beyond “text in, text out.” Today’s AI is increasingly:
- Multimodal: It can work across text, voice, images, and sometimes video.
- Long-context: Modern models can handle extremely large context windows, which changes what’s possible in knowledge work.
- Infrastructure-like: Foundation models are increasingly treated like a base layer that other products build on.
- More real-time and more on-device: Voice and assistant-style experiences are becoming more seamless and more natural.
This matters because it changes the GTM opportunity surface: AI can increasingly “see” context, reason over larger information, and produce outputs that feel immediately usable - if the underlying inputs and rules are strong.
Real-world usage: what AI is being used for in GTM
At a practical GTM level, AI is being used to:
- Draft and rewrite content (emails, messaging, summaries).
- Summarize calls, surface key insights, and reduce manual admin.
- Accelerate analysis by querying data in natural language and compressing time-to-insight.
Whether that actually helps depends on maturity. We’ll come back to that later in the class.
Where AI is heading (near future trends)
Some useful near-future patterns:
- Democratized AI creation: More no-code/low-code AI tools, more “build your own” model paths, more accessible infrastructure.
- Multimodal as the standard: Using mixed media becomes normal, not special.
- Specialized hardware: Continued progress in compute approaches (including highly specialized hardware).
- Agent ecosystems: More systems that behave like teams of assistants coordinating work.
The key is not to get overwhelmed. You don’t need to chase everything - just keep up with the fundamentals and apply the course framework to your business.
Limitations you must understand (so you don’t overtrust)
This is where most teams get burned: overconfidence without understanding constraints.
Some core limitations:
- Lack of continuous learning: Models don’t “learn” like humans through ongoing lived experience. They are trained, then largely fixed, with workarounds used to simulate freshness.
- Disconnected learning: Models train on fragments, not a continuous experience stream.
- Hallucinations and forgetting: AI can produce fluent answers that are wrong, and it can do so confidently.
- No real understanding: AI can describe concepts without genuinely experiencing or comprehending them.
- Scalability and cost: Powerful AI requires significant compute and cost tradeoffs.
There are also societal/ethical risks, including misuse and harmful applications. AI capability is not inherently “good.” Outcomes depend on the people building it and the people using it.
Closing takeaway
AI has evolved from narrow, single-task systems into increasingly multimodal engines that can see, hear, and reason across context. Yet it still lacks true understanding, continuous learning, and clean ethical balance. The next breakthrough must focus on humanity as much as capability.
