nigel's blog

In our daily lives, we make countless decisions, unaware that each choice is a gamble of its own. Whether it’s a bug fix, a new feature, or a product requirement, every decision we make carries varying levels of risk and reward. As professionals involved in product development, it’s our responsibility to make informed choices that drive progress. But do we truly understand the stakes we’re playing with?

Instead of avoiding or neglecting the inherent gambles in our decision-making processes, it’s time to consciously acknowledge them. Let’s challenge ourselves by asking, “How much am I willing to wager on this decision?” This is not about assigning blame; it’s about embracing the accountability of our choices and striving for better outcomes.

To make this decision-making process more effective, consider finding a “gambling buddy” – an accountability partner who shares your commitment to making thoughtful choices. Together, you can create a shared document, perhaps a Google doc, where you outline the problem at hand, propose a hypothesis, and define a success metric. The success metric is the bet, call your shot, how effective will your change be? Only hard numbers allowed in the success metric.

Wager Formula
  Problem: "Not enough users signing up"
  Hypothesis: "Move sign-up button to top of the page"
  Time: 2 day construction. Test: 15 days
  Success Metric: 30% conversion rate

Here’s a lightweight Google Doc Template of this formula, feel free to comment for changes or use as you need.

An example of my usage of the bet, came during a planning exercise. After uncovering a high pain point with a fast solve I proposed the solution. This came into direct conflict with our current work stream, so I have to work with the cross-functional team to evaluate. We work through the details, and I go back to my micro sounding board team to gamble. How much am I willing to bet this deviation in the work stream is faster, and higher value than current efforts?

Businesses, like buildings, require a foundation. For a building, stone and thick concrete are proven stable foundations. In business, any item that helps and holds high volumes of people with decreasing noticeable pain. Examples of this are, telephone infrastructure, powerthat people funnel into are stable foundations. Examples of this are telephone, power, water, and rail networks.

Networks like telephone or rail have become invisible to the everyday user, we interact with the systems built on top of them. For telephone a basic example of this progression is telephone line > internet > network servers > Facebook > Local Marketplace. In this flow most users connect to Facebook > Marketplace and the lower, foundational, layers of the stack are invisible. Think about those lower layers, they operate 24/7 with extremely high confidence of longevity because of the number and volume of uses above them. That level of operation is business goal and benchmark I use when looking at technology.

We can use the progression of foundational technologies—such as the progression from phone lines, to the internet, to connected servers, and eventually to platforms like Facebook—provides a framework for assessment. Several questions arise when judging the potential and effectiveness of a new technology: How many people can this technology support? How challenging is it to acquire new users? What is the duration of each user’s engagement on the platform (time spent)? Does the platform encourage users to invite others (virality)? And, how often do users return to the platform (user retention)?

By addressing these questions, we can better understand the potential impact and longevity of new technologies.

Stay tuned for more on this topic. We will use this model to review modern technology in AI.

Cheers,

Nigel A.

Assumption: AI systems share data vs Google Search/Wikipedia/Blog post

Why would a user choose AI over current data/learning systems?

1. Speed. Can we communicate with AI more easily than google search to answer a question?
2. Unbiased data. Can AI give me a clean answer to my question without tampering with the results?
3. Comprehensive. The answer is more complete from a single source vs compiling results from various sources.

AI Answer

1. Personalization

• AI: Can provide personalized recommendations and data based on user behavior, preferences, and historical interactions.
• Traditional Systems: Offer static information that requires user navigation and interpretation.

2. Proactive Assistance

• AI: Predicts user needs and proactively provides relevant information and assistance without explicit queries.
• Traditional Systems: Generally, react to user inputs and do not anticipate user needs.

3. Dynamic Learning and Adaptation

• AI: Learns and adapts to the user’s evolving needs and preferences, continuously improving its interactions and recommendations.
• Traditional Systems: Do not adapt or learn from user interactions and provide the same experience to all users.

4. Accessibility and Inclusivity

• AI: Can provide more accessible and inclusive experiences, adapting to different user needs (e.g., voice commands for visually impaired users).
• Traditional Systems: May not cater to the specific accessibility needs of diverse users.

---

Trade-offs

Are there trade-offs to be considered before or while using AI as your primary source of knowledge? Always, and when embarking on a journey its in our best interest to consider them before proceeding.

1. Privacy and Data Security Concerns

• Detail: AI systems often require access to personal data to deliver personalized experiences, which might raise concerns about data privacy and security.
• Implication: Users might be reluctant to share personal information, and there is a risk of data breaches or misuse, potentially harming individuals and diminishing trust in AI systems.

2. Potential for Bias and Inaccuracy

• Detail: AI systems can inherit biases present in their training data or algorithms, which might result in unfair or inaccurate recommendations and decisions.
• Implication: This could lead to perpetuating stereotypes, providing skewed information, or making unjust decisions, which might disadvantage certain user groups.

3. Over-reliance on Technology

• Detail: As AI systems take over more functions, there’s a risk of users becoming overly reliant on technology for information, decisions, and tasks.
• Implication: This could potentially diminish critical thinking skills, reduce human oversight in decision-making, and make users vulnerable to errors or limitations of the AI.

4. Accessibility and Digital Divide

• Detail: Advanced AI functionalities might require sophisticated devices and reliable internet access, which may not be available to all user demographics.
• Implication: This can exacerbate the digital divide, where individuals without access to advanced technology are disadvantaged and unable to benefit from AI-enhanced systems.

Where do we go from here?

This series is designed to open experiments and test AI. In this case, we will experiment with AI as a primary education tool and survey the landscape. As a creator, it is exciting to think of the potential in building a tool that will allow for potentially non-biased, fast, accurate, tailored data. My presumption is this is a crowded field with many candidates vying for position, including household tech names.

Along with the high potential for competition, revenue creation is unclear as a data retrieval and creation tool needs to target a wide audience which can/should lower price to entry. I believe, we grow as we broaden educational access, seeing this space go to zero would be an amazing outcome for society.

Stay tuned for the experiment and outcome

Cheers

Nigel A.