What Constitutes a Product?

The definition of a "product" might seem straightforward, yet it remains a topic of debate within the product community. Some refer to it as a "digital product," emphasizing its digital nature as a contrast to physical goods. Others adopt a more philosophical approach, describing a product as an iterative, never-obsolete unit of sale. According to Cagan's 2019 article, a clear and concise definition is:

Product = Customer x Business x Technology

Breaking this down:

Technology

When we think of a product, technology often comes to mind first. In the era of the internet, successful products have largely been tied to disruptive technologies such as Netflix, the iPhone, Instagram, and Amazon. Technology serves as the backbone of what transforms an idea into a marketable product. It is the tangible aspect that captures our attention, much like Edison's light bulb.

Business

In contrast, business encompasses all elements that elevate the technology to its full potential—sales, management, marketing, and finance. While it may lack the allure of groundbreaking technology, the business aspect is foundational to profitability. The primary goal remains unchanged: generate profit. A company’s effectiveness in balancing revenue growth and cost control is what ultimately strengthens its business.

Customer

A successful product requires a business to sustain it, and a business exists only if there are customers willing to pay for the technology. The focus on customer-centric strategies is not a new concept in UX or Product Management; however, the shift in perception from viewing customers as mere targets for exploitation to recognizing them as integral to strategic planning is significant. Without customer funds, a business cannot thrive, and without addressing customer needs, relevant technology cannot emerge.

Challenges Facing Generative AI in Productization

The question of whether generative AI can be classified as a product might seem simplistic, but applying Cagan's equation reveals potential shortcomings for figures like Sam Altman. The technology is undoubtedly impressive; innovations like ChatGPT represent some of the most exciting advancements since the internet's inception.

However, the critical elements of customer base and business model appear to be lacking. OpenAI charges for access to ChatGPT-4 and its API, yet there are concerns regarding profitability given the plethora of open-source alternatives and the high costs associated with training models. Many speculate that they may be reliant on investor funding rather than sustainable revenue.

The situation is akin to Edison's struggles during the battle for electrical supremacy—where cheaper, scalable alternatives posed a threat. If OpenAI and other AI enterprises cannot differentiate themselves from free options, they may find it challenging to establish a profitable business model.

The streaming industry, for example, overcame piracy not by offering more free content, but by enhancing the user experience for on-demand viewing. Similarly, AI companies may need to prioritize user experience over technology to carve a niche in the open-source landscape. However, identifying a suitable target audience is a complex task, as diverse user needs complicate the development of a universally appealing solution.

From a technical standpoint, the industry faces additional hurdles. As AI becomes increasingly prevalent, a recent study raises concerns about training algorithms on outputs generated by other algorithms. This could lead to a saturation of AI-generated content, potentially stifling future AI advancements.

Can We Productize Generative AI?

As a product manager, my role involves developing and planning products. If I were to approach the productization of generative AI, I would start by addressing five key questions:

1. Does the user want this?
2. Can we build it?
3. Are users equipped to utilize it?
4. Is it financially viable?
5. Is it ethically sound?

Based on my analysis, it is evident that we can build generative AI and that there is significant user demand, as demonstrated by ChatGPT's rapid growth to 100 million monthly users within just two months.

However, the long-term challenges lie in monetization and ensuring that users understand how to effectively use the product. We need to maximize value for our users while also generating revenue.

Two notable companies that have succeeded despite initially vague end-user definitions are Meta and Google. Both of these tech giants initially offered broad-use applications that have since become integral to daily life. Their success can be attributed to discovering that a byproduct of their large user bases was digital advertising revenue—a concept that many once deemed impossible in a decentralized online environment.

To summarize, yes, we can productize AI, but the timing may not be right.

Transforming AI into a Marketable Product

If you were hoping to find a definitive answer on how to convert generative AI into a product, I must disappoint you. If I had a surefire method, I'd be cashing in a lucrative deal with Sam Altman rather than writing this post.

While I don’t have a clear-cut solution, I do have some ideas based on the challenges and historical precedents discussed earlier.

AI Advertising Business

Currently, this concept is not feasible, as prioritizing one output over another based on arbitrary ad investments contradicts the fundamental principles by which AI algorithms are developed and assessed. However, it’s likely that someone is exploring ways to make this technically viable even as you read this.

Companies like Google and Meta have largely relied on advertising as their revenue source. Google is notorious for its graveyard of discontinued features, while Meta has faced criticism over its metaverse ambitions. This points to the idea that monetizing extensive user bases may hinge on a model similar to digital advertising.

Human-Curated Training

This idea builds on the earlier discussion regarding the risks of AI algorithms relying solely on outputs generated by other AIs. As the internet becomes increasingly saturated with AI-generated content, the value of human-created content could skyrocket.

Establishing a system to differentiate human from AI-generated content at scale presents significant challenges. However, a future may emerge where AI companies charge a premium for human-generated content, akin to current practices where social media accounts created by humans are valued higher than their bot-created counterparts.

AI as a Service (AIaaS)

Transitioning from general-purpose AI to specialized, user-friendly applications could significantly enhance user experience. Companies like OpenAI and Google could focus on delivering tailored AI solutions for specific use cases, which would not only improve market appeal but also create new revenue streams through hosting or transaction fees charged to businesses utilizing their models.

While ethical concerns such as data privacy and potential monopolies remain, this approach offers a more optimistic outlook for the future of generative AI, highlighting the importance of user experience while harnessing its transformative capabilities.

Harnessing the Power of AI in Product Management - This video delves into how AI can be effectively integrated into product management strategies.

A Day in The Life of An AI Product Manager - Explore the daily responsibilities and challenges faced by product managers in the AI industry.