AI amplifies Cognitive bias

What is cognitive bias?

Cognitive biases are systematic patterns of deviation from norm or rationality in judgment that lead to individuals creating their own "subjective reality" from their perception of the input. These biases often result from our brain's attempt to simplify information processing, leading to errors in reasoning and decision-making. These biases are are often influenced by past experiences.

The intent of this blog is not to nudge you to remember different types of cognitive biases and the scientific terminologies, rather we want you to come to a self-realisation that one's individual opinions may not be always right and it is a good practice to gather inputs and decide as a team when you are designing a product. Always, always test your assumptions with the customers or targeted personas and be open to pivot or course correct when you find them to be proved wrong.

You are a bouquet of your past

You are a bouquet of your past — every decision, every emotion, every shortcut your mind has taken to make sense of the world still lives within you. These arrangements are called cognitive biases, and they silently color how we see and choose.

Memory and Availability
We tend to rely on information that is most readily available to us, often from recent experiences or memorable events. This can lead to the availability heuristic, where more recent or vivid memories disproportionately influence our decisions.

Learning and Conditioning
Past experiences teach us what to expect in certain situations. This learning can lead to biases such as confirmation bias, where we seek out information that confirms our previous beliefs and ignore information that contradicts them.

Patterns and Expectations
Our brains are wired to recognize patterns and make predictions based on them. When faced with new information, we often fit it into existing frameworks formed by past experiences, which can result in biases like the anchoring effect, where initial information serves as a reference point for all subsequent decisions.

Emotional Associations
Emotional experiences leave strong impressions and can heavily influence future judgments and decisions. This can lead to biases such as the affect heuristic, where emotions play a central role in decision-making, often at the expense of rationality.

Social Influences
Social and cultural experiences shape our perceptions and biases. Groupthink and the bandwagon effect are examples where social dynamics and past social interactions influence our thinking and decisions.

Lets classify our cognitive biases

Confirmation Bias

• Description: The tendency to search for, interpret and remember information that confirms preexisting beliefs.

• Impact: Teams may focus on feedback that supports their initial ideas while ignoring critical feedback, leading to products that do not truly meet user needs.

Anchoring Bias

• Description: The tendency to rely heavily on the first piece of information encountered (the "anchor") when making decisions.

• Impact: Initial estimates or first ideas can unduly influence the direction of the product, leading to a lack of consideration for alternative solutions.

Availability Heuristic

• Description: Overestimating the importance of information that is readily available or recent.

• Impact: Decisions may be based on recent customer feedback or visible market trends, overlooking broader or long-term data.

Overconfidence Bias

• Description: Excessive confidence in one’s own answers or predictions.

• Impact: Teams might underestimate risks, overestimate user acceptance, or believe too strongly in the success of a feature without adequate testing.

Bandwagon Effect

• Description: The tendency to do (or believe) things because many other people do (or believe) the same.

• Impact: Following popular trends without critical evaluation can lead to products that lack unique value propositions or fail to address specific user needs.

Sunk Cost Fallacy

• Description: The inclination to continue an endeavor even with little to no success once an investment in money, effort or time has been made.

• Impact: Teams may persist with failing projects or features because of the resources already invested, rather than cutting losses and pivoting.

Recency Effect

• Description: The tendency to weigh recent information more heavily than older data.

• Impact: Recent user feedback or market changes might disproportionately influence product decisions, overshadowing established research and data.

Survivorship Bias

• Description: Concentrating on the people or things that "survived" some process and overlooking those that did not because of their lack of visibility.

• Impact: Focusing on successful features or products without considering those that failed can lead to overestimating the likelihood of success. This bias is particularly dangerous to start up founders.

Planning Fallacy

• Description: Underestimating the time, costs and risks of future actions and overestimating the benefits.

• Impact: Product development timelines may be overly optimistic, leading to delays and cost overruns.

Groupthink

• Description: The desire for harmony or conformity in the group results in an irrational or dysfunctional decision-making outcome.

• Impact: Teams may avoid conflict and dissenting opinions, leading to a lack of critical evaluation and innovation. One of the many reasons why Brainstorming sessions are less successul.

So how can we avoid it?

Foster Diversity and Inclusion
Build teams with diverse backgrounds to challenge assumptions and provide varied perspectives.

Emphasize Data-Driven Decisions
Base decisions on robust data and analytics to ensure objectivity and accuracy.

Implement Regular Feedback Loops
Establish continuous feedback from users and stakeholders to validate assumptions and decisions.

Encourage Constructive Dissent
Create a safe environment for team members to voice dissenting opinions and challenge the status quo.

Utilize Structured Decision-Making Frameworks
Employ systematic approaches like SWOT analysis and decision trees to evaluate options and mitigate biases.

Using AI to Counter unconscious bias

Cognitive bias can’t be eliminated, but it can be managed through awareness, critical reflection, and seeking diverse perspectives. In the age of AI, this becomes vital: biases don’t just affect individual choices, they ripple into data, models, and algorithms that scale decisions across millions. If unchecked, our hidden distortions risk being amplified into systemic flaws. By actively questioning assumptions and embracing diversity, we ensure that both humans and machines serve fairness, clarity, and truth—rather than unconsciously reinforcing yesterday’s blind spots.

Use AI to counter unconscious bias (source: IxDF)
Companies have developed algorithms to help detect biases in AI. Below are some examples. Each of these tools approaches the question of ethics from a different perspective:

• Fairlearn: An open source toolkit from Microsoft for data scientists and developers to assess and improve the fairness of their AI systems.

• Accenture’s AI testing services: Rely on a “Teach and Test” methodology to train AI systems to avoid biases.

• Bias Analyzer: A cloud-based application by PwC that flags potential biases in AI outputs.

• FairML: A toolbox developed by MIT student Julius Adebayo for auditing predictive models by analyzing the model's inputs.

• Google's What-If tool: This tool questions what fairness means and allows product developers to sort the data according to a different type of fairness, allowing humans to see the trade-offs in different ways to measure fairness and make decisions accordingly.