Anyone who says they’re an AI expert is lying

Some of the brightest brains in tech are not AI experts yet.

I left Spark Accelerate last week with my head hurting from trying to retain all the information generated by the incredible speakers and content. One of the standouts was Edmundo Ortega’s keynote, “An Executives Guide To Building AI For Enterprise.” Ed and I have become friends, so I must admit a bias when I say this…but the data is clear. Everyone I spoke to from the attendees found his talk illuminating and insightful.

With a week to go back to everyday life and time to review both the shaky images taken of Ed’s slides I took and my barely decipherable notes in my notes app, I will do my best to recap some of the key highlights and takeaways.

Welcome to Aotearoa, Eeiid

Edmundo Ortega, or as we Kiwis (with our accent) call him, “Eeiid,” is a partner at Machine & Partners, a strategic consultancy that helps organisations get up to speed and into the market with AI products that deliver real and relevant value. He’s also a lecturer at Section School, the premier AI upskilling platform (I’m the ANZ ambassador for Section).

Ed was only here briefly, but I think he experienced enough of Tāmaki Makarau (Auckland), thanks to his hosts at Spark and all the incredible people he met, to realise how cool it is here. For my part, I got to take Ed to some great spots, including Piha’s incredible Mercer Bay Loop Walk, thanks to a tip from our fantastic friend Janeen Anyon.

Anyone who says that they’re an AI Expert is lying.

Ed’s provocative statement opened early on in his talk. Ed believes that not only is he not an AI expert and doubts anyone can be an expert at this time. Instead, Ed and his amazing cohort at Machine & Partners apply best-in-breed product strategies and combine them with a creative, process-driven innovation approach to harness this powerful new technology.

AI is both overhyped and not-hyped enough

“AI is overhyped. And it’s going to change how products are built and delivered fundamentally.” Edmundo Ortega

It’s an impressive technology that is easy to start with and build “toys” (novel tools that don’t shift the dial for the business) with but hard to get right. In fact, according to Gartner, 30% of all generative AI projects in 2024 will not go beyond the proof-of-concept stage.

The types of AI you need to know.

One sign I am not an AI expert is that I hadn’t heard of at least one of the models that make up AI. They are:

• Machine Learning (ML): Ideal for handling vast amounts of data to make predictions, like optimising route planning for city services.
• Large Language Models (LLM): These models are impressive but struggle with factual accuracy and reasoning, making them better suited for creative tasks than precise data-driven needs.
• Retrieval-Augmented Generation (RAG): A hybrid approach that combines LLMs with proprietary data to provide accurate, context-specific answers.
• AI Agents are powerful tools that offer fine-grained logic control and decomposable functionality.

How about DEW?

But how do you identify possible use cases for AI? One way of looking at where AI best plays is the DEW framework, which looks for human processes that AI can enhance or replace.

• Data: AI is great for analysing and interpreting structured and unstructured data, including text and images.
• Expertise: we’ve all got those people in our businesses who have a scary amount of knowledge and experience in their heads. AI presents an opportunity to enhance their experience while minimising risk from their absence.
• Workflows: dreary, manual, repetitive tasks are great candidates for AI automation. But even advanced use cases are in play for AI.

It’s all about SME

AI is a different technology from its predecessors—so much so that even the way we work with it needs to change. AI product teams should be created with the subject matter expert at the forefront and integral to the team. Technologists and product managers are also vital, but AI’s nature means that the SME must heavily inform their work.

AI adoption agency

Exploring how enterprises adopt AI at a reasonable scale and pace requires one of three approaches: “narrow”—key people, leaders, and AI gurus; “wide”—made available widely for everyone to use, creating use cases; “deep” – a combination of both, wide adoption with AI gurus cultivated to lead and guide.

Top tips for AI product development

• AI policy: have one, and don’t make it too restrictive, encourage innovation 
• Product strategy: choose your projects carefully but stay ambitious
• Innovate now, integrate later: don’t integrate AI immediately; build it “off to the side”.
  

If not an AI expert, then Ed is pretty close.

Ed’s keynote was a refreshingly honest take on where AI stands today. To win, you need a practical, curious, and collaborative approach, which describes Kiwi innovation and strategy at our best.

Ngā mihi nui Ed! We appreciate you. Thank you to Spark – an inspiring event.

A lot to think about – what next?

Ed’s LinkedIn is here. If you’re interested in AI adoption in your business, I can help by connecting you with Section; if you need performance marketing with AI and innovation at the core, Europa Creative Partners can help. Either way, get in touch—I would love to hear from you.