What AI can and can't do right now

In boardrooms and startup hubs worldwide, generative AI is being hailed as the next industrial revolution, promising to automate creativity, personalize experiences, and slash costs overnight. Yet as tools like advanced language models flood the market, executives face a deluge of promises that often outpace delivery, risking wasted investments and missed opportunities. This tension makes understanding AI's true boundaries essential for any business leader.

The Speaker's Lens on AI Capabilities

Andrew Ng's talk "What AI can and can't do right now" offers a grounded lens for examining generative AI's role in business. Ng breaks down current capabilities of AI systems, including generative models, while highlighting their strategic implications. His core thesis stresses that businesses unlock value by focusing on realistic applications and steering clear of common pitfalls tied to overestimating what these systems can achieve today.

Balancing Hype with Practical Limits

Ng's ideas reinforce the trending conversation by affirming that generative AI holds strategic promise when deployed with clear-eyed awareness of its boundaries. They complicate the narrative by reminding leaders that hype around limitless potential can obscure operational constraints, leading to misaligned strategies. This reframes the central question of balancing hype with limits: rather than chasing every new model, companies must map AI tools directly to measurable business outcomes.

• Prioritize use cases where current generative capabilities align with existing data strengths.
• Audit projects for hidden dependencies on human oversight that hype often ignores.

Driving Real Business Value

By synthesizing these points, Ng equips organizations to move beyond broad enthusiasm toward targeted experimentation. His breakdown of capabilities and limits provides a filter for evaluating whether a generative AI initiative will deliver tangible returns or simply inflate expectations.

What one generative AI pilot in your organization will you rigorously test against its actual capabilities this quarter?