78: Juan Mendoza: The ethics of generative AI, trust, transparency and the threat of dehumanization

What’s up everyone, today we have the pleasure of chatting with the profoundly eloquent Juan Mendoza.

Summary: Juan unpacks the unique journey of OpenAI, underscoring the importance of distinguishing between tech hype and real value. As AI intertwines with our lives, Juan highlights the delicate dance of harnessing its efficiency while preserving human creativity. He calls for a critical balance in using AI as a creative aid without stunting our own creative prowess. Amid the benefits, Juan also raises crucial questions about trust and privacy, advocating for marketers to use AI responsibly. Ultimately, this episode reaffirms the need to thoughtfully navigate AI’s limitless potential while upholding our fundamental human values and ethics.

About Juan

• Juan is based in Melbourne, Australia where he got his start in various roles including startup marketing, martech strategy and conversion rate optimization.  
• He spent 4 years at The Lumery, a premier Martech consulting shop where he worked on customer centric strategies across various channels.
• In 2020, he started the Martech Weekly, a newsletter covering where the industry is going and why.
• He’s also the host of the Making Sense of Martech podcast, an extension of his newsletter
• One podcast and a newsletter isn’t enough for Juan though, in 2022 he teamed up with Scott Brinker and started the Big Martech show covering big news and big ideas in Martech
• 4 months after launching a premium subscription and growing his newsletter to thousands of martech pros over 65 countries, Juan decided to go full-time on TMW 
• Recently he announced TMW 100 – a global Martech awards event ranking the most innovative marketing technology companies from 1st to 100th place

The OpenAI Approach: Laser-Focus and High-Quality Language Models

Juan underscored the significance of OpenAI’s niche focus and dedication to developing large-scale language models. Unlike big tech giants—Google, Amazon, Meta and others—who spread their resources and attention over various types of AI technologies, OpenAI chose a distinct path. They concentrated all their efforts on building transformative generative models like GPT-3 and GPT-4, which set the groundwork for the success of ChatGPT.

Juan suggested that OpenAI’s edge lies in their extreme focus, patience, and funding. Established as a Silicon Valley tech company, OpenAI was backed by tech tycoons like Elon Musk and Microsoft, as well as some of the world’s largest venture capital firms. However, OpenAI’s goal differed. They weren’t trying to distribute AI across various services and products. Instead, they strived to create something unique and powerful—a tool that could manipulate language with an unprecedented level of precision.

Juan further highlighted how OpenAI’s technology feels “magical” compared to competitors. For instance, Google’s BERT, although an impressive model, doesn’t meet the same level of accuracy as ChatGPT. Moreover, it ‘hallucinates’—generates incorrect or nonsensical outputs—significantly more often than ChatGPT.

Juan also reminded us of the chatbot craze of 2016, which, despite the hype, resulted in less-than-stellar customer experiences. Fast forward to today, the launch of GPT-based models has reinvigorated the chatbot space, breathing new life into the industry. The key difference? An AI agent that can actually provide intelligent, useful responses.

Takeaway: OpenAI’s specific concentration on creating large, high-quality language models, backed by extreme patience and funding, was pivotal in their success story. Their dedication to a niche allowed them to develop an AI that is more accurate and less likely to hallucinate than its competitors. This focus transformed the realm of chat interfaces, redefining the future of AI agents and encouraging a fresh wave of startups to build on this advanced technology.

The Adoption Race: Comparing ChatGPT and Snapchat’s AI Tools 

Juan’s analysis of the rapid spread of ChatGPT brought fascinating insights. He cited how ChatGPT became one of the fastest-growing apps globally, reaching a million users in merely five days. This impressive rate outpaced even Instagram and TikTok’s growth, both of which took weeks to months to reach the same number of users. He attributed this meteoric rise to the simple yet brilliant user experience, which only required users to create an account and start chatting.

However, Juan posed an interesting counter-argument, comparing ChatGPT’s growth with the newly launched AI tool by Snapchat—Snap AI. Despite being on the market for only two months, Snap AI already boasts 125 million users and over 10 billion messages sent. It reached 100 million users faster than ChatGPT, achieving the feat in just two months compared to ChatGPT’s four to five months.

Juan speculated on the factors driving Snap AI’s rapid growth, posing two possibilities. One might be the timing—Snap AI could be riding the wave of excitement and acceptance generated by GPT-based technology. Another potential factor could be a generational shift, with younger users flocking to the more AI-enhanced environments provided by platforms like Snapchat.

Takeaway: While ChatGPT has been a massive success, becoming one of the fastest-growing apps globally, other AI tools, like Snap AI, are rising quickly in the adoption race. The rapid growth of these platforms may be due to a combination of riding the AI hype cycle and a generational shift towards AI-enhanced environments. It is a reminder that, in the ever-evolving AI space, being first to market doesn’t always mean you’ll be the fastest-growing or most widely adopted solution.

The Metaverse, AI and the Hype Cycle: A Critical Analysis

Juan expressed his concerns about the hype cycle surrounding emerging technologies like AI, the metaverse, and web 3.0. He opined that such hype cycles often lead to exhaustion, especially among senior individuals who are tired of constant pitches for the “next big thing.”

Juan cited an example where he was at a conference and representatives from Meta and an advertising agency were zealously promoting the metaverse. According to him, they were pushing marketers to prepare for the metaverse, without critically analyzing why it should be included in their marketing strategies in the first place.

Juan also expressed disappointment in certain consultancy companies for their uncritical acceptance of Mark Zuckerberg’s vision for the metaverse. McKinsey, Accenture, and Bloomberg were among those criticized by Juan for their generous predictions of the metaverse’s economic impact, without rigorous analysis of the feasibility and practical implementation of this new trend.

Contrasting this with the iPhone, Juan pointed out that it took five years for the iPhone to reach mass consumer market penetration, despite it being recognized as one of the most significant shifts in consumer experiences, software development, and mobile technology. Thus, according to Juan, hype without actual products often...