Navigating Probability-Infested Times

The Illusion of Intelligence

We are living in probability-infested times.

Text models predict the next word. Image models fill in noise. Code models autocomplete syntax. And all of them do so without true understanding.

This doesn’t just lead to hallucinated facts; It leads to hallucinated decisions, with very real consequences.

What We Will Explore

Generative AI is the new kid in town and everyday, we hear more people claiming years and years of experience in it (as well as tons of experience in Agentic AI and god knows what else). Anyway, it is not possible to ignore the beast so, that alone means it deserves its own category. Having said that, let me be clear that, in the right hands (or as those in the hood call it, with the right prompt) it can be a wonderful tool that can massively boost our productivity.

I am actually a big fan, as you will see throughout the coming posts, but, I refuse to ignore its deficiencies or sell perceived capabilities that are simply not there yet (and some of those capabilities are unlikely to ever be there as those are tightly coupled with the initial assumptions and the overall approach). Think of it this way, if real AI (or AGI as some will call it) is compared to Relativity, then the current Generative version of AI is equivalent to Newtonian Mechanics (I love Newton, but there are problems that his wonderful laws of Mechanics are just not equipped to answer, hence the birth of Relativity and the emergence of a certain Albert Einstein).

Now, enough of this. For this introductory post, let’s say that under Generative AI category, we will cover many topics including the following:

Why relying exclusively on probabilistic logic is both risky and regressive,
What AI Agents vs. Agentic AI really means.
Silly but illuminating examples of probabilistic failure in daily life that can help in our overall understanding of Generative AI, its strengths and its limitations,
How we might use Quantum logic, structured data, and human values to rebuild trust.

This space is crowded with hype. Our aim here is to teach clearly, challenge assumptions, and reclaim agency.

The above tagline was originally written as “Our aim here is to cut through the hype and some monumental BS so that we can better understand our assumptions and see more clearly.” The end result, however, is a revision by ChatGPT:-). I think we can all agree that ChatGPT’s output is more succinct and certainly more polite so, hats off to ChatGPT here. Like I said, if you know how to use these toys, they find a way to please you.

Related posts:
Quantum Computing | AI Ethics | The Hidden Thread

What CEOs, Lawyers, Politicians, and Journalists Have in Common

Good day, my good readers!

I woke up with a headache today (it happens frequently, specially when I go to bed reading about the world events) and immediately found myself wrestling with a dirty the following “dirty” thoughts:

When we think of CEOs, lawyers, politicians, and journalists (or the Painful Four, as some may call them), we usually imagine vastly different worlds: boardrooms, courtrooms, parliaments, and press rooms. But beneath the surface, these professions often operate with surprisingly similar habits:

1- They tend to prefer binary answers in a world full of ambiguity (as a trained mathematician, physicist and computer scientist, working for more than 2 decades on the latest technologies, I have a healthy respect for the binary so, just hear me out, OK?).

2- They look for quick fixes rather than long-term systemic change.

3- They often rely on technical tricks or simplified narratives to gain leverage.

4- And sometimes, they rely on drama to get attention.

At their best, these habits can be used for real public good. At their worst, they amplify division, mask complexity, and concentrate value among a small few. The rest of the society is left with oversimplified stories and underwhelming outcomes. This becomes especially problematic (even dangerous) when these patterns affect how we use and regulate powerful new technologies such as Generative AI (that is for another post to delve into under Generative AI category. I started this post as a Rant so, forgive me for continuing the rant, I have a headache, remember?).

Now, let’s take a closer look.

CEOs: Impact Versus Optics

The Best Case
Yvon Chouinard, founder of Patagonia, transferred ownership of the company to a trust and nonprofit. His goal: ensure that all future profits are used to combat climate change and protect the planet. It turns out that, this was not a branding stunt or a tax dodge. It was a rejection of short-term profit-taking in favor of long-term environmental responsibility.

The Worst Case
Elizabeth Holmes promised revolutionary medical diagnostics. Instead, she sold a black-box illusion that fooled even experienced investors. The result was broken trust and burned billions.

The Pattern
We reward CEOs who pitch bold solutions, especially if they sound futuristic or, worse, unrealistic. We are seeing this “Impact vs Optics” dilemma playing itself out with the surge of Generative AI. Generative AI is based on probability and pattern prediction, not deterministic or logical approach. When we force complexity into simple business narratives, we end up creating tools and results that are misunderstood, misapplied, or mis-sold.

Lawyers: Justice Versus Justification

The Best Case
Bryan Stevenson, founder of the Equal Justice Initiative, uses the law to challenge systemic injustice. His work is slow, complex, and grounded in ethics.

The Worst Case
Legal teams working for Big Tobacco and Big Oil have spent decades defending harmful practices through legal loopholes. They win cases by focusing on technicalities, not on harm reduction. I often find myself admiring lawyers for their innovative work with the law, and the language, when using a nasty loophole, while, at the same time, feel bitterly disappointed at watching them justify the harm their brilliant legal maneuvering has done to their victims.

The Pattern
Legal culture often celebrates the win, not the impact. In the world of AI, where systems are embedded into compliance processes and automated decision-making, legal professionals may hide behind technical accuracy. But law, like code, reflects values. If ethics are not baked in, then technical correctness becomes a shield for irresponsible systems.

Politicians: Substance Versus Soundbites

The Best Case
Angela Merkel often took political risks to communicate complex decisions with clarity. Her leadership was not perfect, but it was thoughtful and consistent. It is hard to find a clear instance that she lied on anything for the sake of electability and that is a lot to say in this day and age.

The Worst Case
UK Brexiteers who used oversimplified slogans like “Get Brexit Done” (or some other version of it) to fuel an outcome that was deeply misunderstood, impractical, and wasteful. The consequences are still unfolding.

The Pattern
Political messaging favors clarity, not complexity. That becomes a problem when the issues are technical and long-term, such as climate change, managing misinformation, or designing resilient infrastructure. Simple slogans rarely make good policy, but here we are, looking to vote in the next (or the same) politicians based on how catchy their latest slogans are.

Journalists: Depth Versus Drama

The Best Case
Maria Ressa exposed authoritarian tactics and online disinformation with courage and depth. Her journalism is committed to complexity and accountability.

The Worst Case
On the opposite end, we have the example of the News of the World phone hacking scandal, where journalists illegally accessed voicemails of public figures and even a murdered teenager. The intrusion gave false hope to the victim’s family and interfered with the police investigation. What was framed as sensational reporting ended up exposing a toxic newsroom culture driven by profit, not public service, and ultimately led to the paper’s shutdown and a national inquiry into media ethics.

The Pattern
In a media economy based on attention, nuance is expensive. It is faster to simplify and it pays to be dramatic. As we all find new ways to get angry at each other, the temptation to blur the line between facts and fiction will grow. If journalists cannot hold the line, the public will not know what to trust.

So Who Is Really Responsible?

That is the hard part.

Is it the professionals? Or is it the rest of us, who reward these behaviors? And don’t forget that before professionals become professionals, they are the rest of us!

We are the ones who share clickbait. We demand certainty. We celebrate confidence even when it lacks substance. We are often too impatient for process, and too distracted for detail.

In a world of Deepfakes, synthetic media, and AI hallucinations, our demand for clarity over complexity is no longer just a personal habit. It is a societal vulnerability, a dangerous weakness that can only be exploited.

If we want better leaders, better laws, better journalism, and better outcomes, we need to become better at embracing complexity. That means making space for slower thinking. That means tolerating uncertainty. That means expecting more, not less, from the people and systems that shape our world.

The challenge is not just technical. It is cultural.

Where have you seen these patterns show up in your field? What would it take to shift the system rather than just the symptoms?

Subscribe to my blog or follow me on LinkedIn to get notified when I start ranting again!

25 Not-To-Dos of Data (Hello Blog World 🎉)

Starting any journey is tough, and more so when it involves data! Since I decided to join the Blog community I have been thinking actively of what to write about and it suddenly dawned on me that maybe highlighting a few Not-To-Dos is an easy way to start this Blog experiment. After all, it is always easier to tell people what not to do even if what not to do is what you actually do yourself (and know you should not do!). Hope you are feeling me. Anyway, here comes the first Blog post on this site so, hope you agree with the contents. Happy to hear from you to improve this and future articles. Oh, and remember, 25 is an arbitrary number. Somewhat round, not too small and not too big but we could go to 50 or more, as what is definitely not in short supply is bad data habits:-).

Finally, this is meant to be a way for me to get back to heavy writing, find like-minded colleagues and get positive feedback to learn. Each item really scratches the surface. I am hoping to pick a few of these short articles in the next few weeks and months to explore further. On the other hand, I have been thinking AI Ethics a lot recently so, maybe that will jump ahead. We will see. Too early to say how but now you know what the “plan” is!

For now, here is a list of 25 “Not-To-Dos” of data that (based on first hand experience over 2 decades) each of which, one way or another, affects effective Data Management:

Neglecting Data Governance
1. What is it: Failing to implement a framework that outlines how data is managed, who owns it, and how it’s used.
2. Potential Impact: Poor data quality, lack of accountability, and compliance issues.
3. Best Practice Fix: Establish a formal data governance framework with clear roles, policies, and ownership for all data assets.
Ignoring Data Quality Management
1. What is it: Not monitoring or enforcing the accuracy, completeness, and reliability of data.
2. Potential Impact: Decisions based on inaccurate data lead to poor outcomes and financial losses.
3. Example: JP Morgan’s London Whale incident where poor data quality in risk models led to $6 billion in losses (Google it!).
4. Best Practice Fix: Implement regular data quality checks and establish clear KPIs around data integrity.
Allowing/Encouraging/Tolerating Siloed Data
1. What is it: Storing data in isolated systems where it cannot be easily accessed or shared.
2. Potential Impact: Difficulty in gaining cross-departmental insights, leading to fragmented decision-making.
3. Best Practice Fix: Break down data silos by using centralized platforms or data lakes that integrate all data sources. Remember a centralized solution is not necessarily physical (hope to touch on this one in more details sooner, rather than later).
Over-Complicating Data Architecture
1. What is it: Building overly complex data pipelines and systems that slow down the processing and decision-making.
2. Potential Impact: Increased operational overhead and slower time to insights.
3. Best Practice Fix: Design data architecture to be scalable and modular, focusing on simplicity and agility.
Relying on Manual Data Processes
1. What is it: Using manual data handling methods such as data entry and validation.
2. Potential Impact: Human errors, inefficiencies, and high labor costs.
3. Example: Just Google for examples of financial services firms who lost millions due to manual data entry errors in transactions.
4. Best Practice Fix: Automate data workflows and validations to reduce human intervention and error.
Lack of Standardization Across Data Sources
1. What is it: Failing to standardize data formats, structures, and naming conventions across systems.
2. Potential Impact: Inconsistent reporting, confusion, and difficulty in combining datasets. And maybe I should really say: too many reports for the same topics! I feel many know exactly what I mean here.
3. Best Practice Fix: Establish data standards and enforce them across all data entry points and sources.
Underinvesting in Data Security
1. What is it: Neglecting to implement proper data security measures such as encryption, access control, and monitoring.
2. Potential Impact: Data breaches, compliance violations, and loss of customer trust.
3. Example: Equifax’s 2017 breach exposed 147 million records due to poor security practices.
4. Best Practice Fix: Implement a multi-layered security approach with encryption, access controls, and real-time monitoring.
Not Defining Clear Data Ownership
1. What is it: Failing to assign responsibility for specific data assets to individuals or departments.
2. Potential Impact: Lack of accountability, leading to mismanagement or neglect of critical data.
3. Best Practice Fix: Assign data owners who are responsible for the quality, usage, and lifecycle of their data.
Skipping Proper Data Documentation
1. What is it: Failing to maintain documentation of data sources, transformations, and governance. This is one place that I go on about the dangers of “Agile Methodology”. Not that it is bad on its own, but documentation is typically the first thing that suffers when you seek agility in an organization!
2. Potential Impact: Difficulties in understanding and trusting the data, leading to errors in decision-making.
3. Best Practice Fix: Maintain comprehensive metadata and data lineage documentation to track the flow of data.
Overlooking Data Privacy Regulations
1. What is it: Ignoring or inadequately adhering to data privacy laws such as GDPR or CCPA.
2. Potential Impact: Heavy fines, reputational damage, and loss of customer trust.
3. Example: Google was fined $57 million for GDPR violations due to improper consent practices. An interesting one to Google about, of course:-).
4. Best Practice Fix: Regularly audit data processes for compliance and update practices according to evolving regulations.
Using Outdated or Unsupported Data Tools
1. What is it: Relying on legacy systems or tools that are no longer maintained or secure.
2. Potential Impact: System failures, security vulnerabilities, and reduced performance.
3. Best Practice Fix: Regularly evaluate and update data infrastructure to ensure tools are modern, secure, and supported.
Failing to Prioritize Data Integration
1. What is it: Not connecting data from different systems, leading to fragmentation and incomplete insights.
2. Potential Impact: Inaccurate reporting, poor customer service, and disjointed operations.
3. Best Practice Fix: Use processes (ETL, etc.), middleware, or other relevant methodologies/technologies to integrate data from disparate sources into a single repository.
Assuming All Data is Valuable
1. What is it: Collecting and storing every piece of data without assessing its relevance or usefulness.
2. Potential Impact: Increased storage costs, complexity in data management, and lower performance.
3. Best Practice Fix: Focus on collecting high-quality, relevant data and regularly purge unnecessary or outdated data.
Not Providing Data Literacy Training
1. What is it: Failing to train employees on how to interpret and use data effectively.
2. Potential Impact: Misinterpretation of data, poor decision-making, and underutilization of data tools.
3. Best Practice Fix: Implement regular data literacy training programs tailored to different levels of expertise within the organization.
Lack of Scalability in Data Infrastructure
1. What is it: Building data systems that cannot grow with increasing data volumes or business needs.
2. Potential Impact: Performance bottlenecks, downtime, and costly system overhauls.
3. Best Practice Fix: Design data systems to scale dynamically by using cloud-native architectures and horizontal scaling techniques.
Not Establishing Clear Data Use Policies
1. What is it: Lacking formal policies on how data should be accessed, used, and shared within the organization.
2. Potential Impact: Data misuse, security breaches, or legal violations.
3. Best Practice Fix: Create comprehensive data use policies that define access levels, usage rules, and protocols for handling sensitive data.
Rushing to Implement AI Without Clean Data
1. What is it: Deploying AI models without ensuring the underlying data is accurate, consistent, and complete.
2. Potential Impact: Poor AI model performance and inaccurate predictions.
3. Example: IBM Watson for Oncology made incorrect treatment recommendations due to poor training data.
4. Best Practice Fix: Focus on data cleansing and quality assurance before feeding data into AI models.
Failing to Align Data Strategy with Business Objectives
1. What is it: Implementing data initiatives without considering their alignment with the organization’s overall goals.
2. Potential Impact: Underutilized systems, wasted resources, and missed business opportunities.
3. Best Practice Fix: Develop a data strategy that directly supports key business objectives and regularly review its effectiveness.
Not Archiving Old or Unused Data
1. What is it: Retaining data indefinitely without determining its relevance or necessity.
2. Potential Impact: Increased storage costs and legal risks from holding onto unnecessary or sensitive data.
3. Best Practice Fix: Implement data lifecycle management policies that archive or delete old data based on usage patterns and regulatory requirements.
Assuming Cloud Migration Solves All Data Problems
1. What is it: Believing that moving data to the cloud will automatically resolve governance, quality, or integration issues.
2. Potential Impact: Continued data problems, only now in a cloud environment, along with unexpected costs.
3. Best Practice Fix: Plan cloud migrations carefully, addressing data governance and quality beforehand, and ensure cloud solutions are cost-effective and scalable.
Lack of Real-Time Data Access
1. What is it: Failing to provide access to real-time data, limiting the ability to respond quickly to changes.
2. Potential Impact: Decision-making based on outdated information, leading to missed opportunities or poor outcomes.
3. Best Practice Fix: Implement real-time data streaming or event-driven architectures to provide up-to-date insights.
Relying Too Much on One Vendor
1. What is it: Becoming overly dependent on a single vendor for critical data infrastructure or tools.
2. Potential Impact: Vendor lock-in, pricing power shifts, and increased risk during outages or failures.
3. Best Practice Fix: Diversify vendors and maintain vendor-neutral solutions where possible, with contingency plans in place.
Over-Reliance on Shadow IT for Data Management
1. What is it: Allowing departments to create and manage their own data solutions outside the oversight of the central IT team.
2. Potential Impact: Uncontrolled data sprawl, security risks, and lack of governance.
3. Best Practice Fix: Integrate shadow IT initiatives into the broader data strategy, providing IT governance while allowing flexibility.
Not Having a Disaster Recovery Plan for Data
1. What is it: Failing to create a backup or disaster recovery strategy for critical data systems.
2. Potential Impact: Data loss, operational disruptions, and severe financial consequences.
3. Example: In 2011, a Japanese automotive company lost critical data due to a natural disaster, as they had no offsite backup.
4. Best Practice Fix: Establish a disaster recovery plan with regular backups (offsite or cloud-based) and conduct recovery drills to ensure business continuity.
Failing to Automate Data Management
1. What is it: Overlooking the automation of core data management functions such as metadata tracking, access control, lineage tracing, and policy enforcement, even when tools exist to streamline them.
2. Potential Impact: Operational bottlenecks, inconsistent governance, reduced data trust, and higher risk of regulatory gaps. Manual oversight simply can’t keep pace with modern data volumes or complexity.
3. Best Practice Fix: Adopt intelligent automation across your data management stack. Automation at the management layer ensures consistency, compliance, and scalability, not just efficiency.

This list is just the beginning. As I hinted at earlier, I plan to explore many of these issues in greater detail over time, and in doing so, I’ll inevitably cover a range of interconnected topics that shape the future of data and intelligent systems.

Here’s what you can expect from my blog posts going forward:

Data Strategy
AI Strategy
AI Ethics
Generative AI
Quantum Computing
Graph Models and Graph Thinking
General Reflections (which this first post falls into)

Whether you’re leading transformation efforts, building intelligent systems, or just trying to make better decisions with data, I hope you’ll find ideas here that challenge, clarify, and contribute.

See you in the next post.