Navigating AI in 2025: Key Takeaways from Tom Gruber (Co-Founder of Siri)

In our recent Demandbase webinar, “Navigating AI Strategies for 2025,” we had the privilege of hosting Tom Gruber, co-founder of Siri and former Head of Advanced Development at Apple, for a timely, wide-ranging conversation about the past, present, and future direction of artificial intelligence.

The discussion, moderated by Elevate’s Chief Innovation Officer Jim Caruso, provided a compelling examination into:

• How AI has evolved over the last decade and is evolving today,
• the opportunity AI represents for today’s businesses, and
• how businesses can use AI responsibly and strategically to support growth.

Summarized below are the key themes and insights from the conversation.

The Evolution of AI: From Siri to Foundation Models

Gruber traced the evolution of AI, starting with voice-activated digital assistant Siri’s early role in creating a widely-used conversational interface. Siri was released as an app for iOS (Apple) in February of 2010. While Siri wasn’t the AI breakthrough itself, it helped establish and define expectations that AI technology, and an AI assistant, could interact intelligently with human users.

The major shift came in the mid-2010s with the rise of deep learning and, crucially, the introduction of transformer architectures in 2017. These breakthroughs enabled AI models to process and learn from massive datasets — essentially all of the internet’s text — without human supervision.

Gruber emphasized the rapid pace of AI advancement, calling this the fastest-moving technology in history, even outpacing Moore’s Law, which says that computing power will roughly double every two years. We’re now living in a world where AI can process the entire internet’s worth of text — literally millions of lifetimes of reading — and generate human-like outputs in seconds.

The acceleration of AI development has resulted in tools that now exceed our ability to fully comprehend or regulate them, raising both opportunities and challenges.

Risks and Misconceptions of AI

AI is powerful, but like any tool, it’s not without its risks. One major concern centers around that old saying from data scientists – “garbage in, garbage out.” If AI systems are trained on unverified or unchecked data (i.e., potential “garbage in”), users face the potential for misinformation and content pollution (i.e., “garbage out”), which could not only lead to massive embarrassment for brands/enterprise users, but also degrade overall trust in AI systems.

Additionally, systems can “hallucinate.” Users often misuse AI as if it were a factual search engine, even though generative models can confidently produce information that’s inaccurate and/or fabricated. As Gruber explains it, “AI doesn’t know truth from fiction. It just generates output in a convincing tone.”

Another misconception is the belief that AI will inevitably replace human workers. Gruber argues that while AI can take over certain tasks (those that are monotonous and repeatable), the most effective use cases for AI are those that enhance human ability, not eliminate it.

A Humanistic Approach to AI

A core theme of the conversation was Gruber’s concept of Humanistic AI — designing AI to amplify human potential rather than eliminate it. This is where B2B marketers can lead the way.

Instead of simply using AI to just churn out blog posts faster, why not use it to personalize content at scale, automate customer segmentation, turn massive intent data into actionable plays, and create smart assistants for sales enablement.

Gruber highlighted mental health care and education as two areas where AI can democratize access and scale expertise. Similarly, AI can empower software development and improve customer service experiences, making it easier for non-experts to create tools or access support more efficiently.

Transparency and Responsibility

Gruber also discussed at length the ethical considerations involved in designing AI interactions with humans. During the development of Siri, for example, the team made a conscious decision not to give the assistant a face or a human-like persona, to ensure that users knew they were interacting with a machine. Humans tend to anthropomorphize the technology they interact with, attributing human characteristics to it, such as agency and/or human emotions.

With today’s capabilities — including hyper-realistic video generation — the line between human and tech is increasingly blurred. Gruber stressed the need for societal norms and legal frameworks to catch up with AI development, so that users are always aware when content or communication is AI-generated rather than human.

In terms of enterprise applications, Gruber described how businesses can get value from general-purpose AI through careful prompt engineering, supplying proprietary data, and designing workflows around specific use cases. For example, running content through various “AI personas” — such as a professor, therapist, or product manager — can yield richer insights and improve the quality of outputs, thus improving a brand’s ability to target and personalize its engagement efforts.

AI’s next evolution isn’t about a single tool replacing your technology stack. It’s about embedding intelligence into every layer of your business, from campaign planning to operational execution to customer success and beyond. As Gruber puts it: “The entrepreneurs — internal or external — who can match business problems [i.e., specific use cases] with AI capabilities are the ones who’ll create the future.”

4 Main Takeaways

What are the most important points for people to take away from this “Navigating AI Strategies for 2025” conversation?

1. AI is no longer experimental or theoretical: it’s here to stay. AI systems are production-ready, and those individuals and organizations who move first will outpace competitors still stuck in “wait and see” mode. As management guru Peter Drucker once said, “the best way to predict the future is to create it.”
2. AI must be implemented with clear guardrails around its limitations and risks. If your AI-generated content is client-facing, for instance, invest in human oversight to ensure that you’re getting the output you want. Use AI to augment human judgment rather than replace it. You can both trust AI systems and also leverage actual human intelligence to verify their output.
3. Start treating prompt crafting like you do copywriting or SEO. It’s an increasingly important skill — and a competitive advantage for individuals and organizations alike.
4. Don’t wait for vendors to define your AI strategy. Identify internal champions, pilot AI across key marketing workflows, and build the muscle to move fast. You know your organization and its needs better than any outside vendor, and you’re the one who should be best positioned to connect AI solutions to your specific use cases.

To learn more about “Navigating AI Strategies for 2025,” watch the full Demandbase webinar.