{"id":1433,"date":"2019-02-15T10:32:09","date_gmt":"2019-02-15T10:32:09","guid":{"rendered":"http:\/\/kusuaks7\/?p=1038"},"modified":"2023-08-23T11:20:14","modified_gmt":"2023-08-23T11:20:14","slug":"why-ai-models-absolutely-need-humans-to-stay-awesome","status":"publish","type":"post","link":"https:\/\/www.experfy.com\/blog\/ai-ml\/why-ai-models-absolutely-need-humans-to-stay-awesome\/","title":{"rendered":"Why AI Models badly need Humans to stay Awesome?"},"content":{"rendered":"

Ready to learn Machine Learning? Browse courses<\/a>\u00a0like\u00a0Robotics Application Machine Learning<\/a> developed by industry thought leaders and Experfy in Harvard Innovation Lab.<\/em><\/strong><\/p>\n

Companies often expect that AI models once implemented will stay\u00a0smart<\/em>\u00a0and work magically, ever\u00a0after<\/h5>\n

Nothing can be farther from the truth.<\/p>\n

<\/p>\n

\u201cWhat about self-healing algorithms or self-maintaining AI platforms<\/em>\u201d you may ask. Well, none are close to production-grade enterprise implementations today. And, it will be a while before they obviate any human intervention.<\/p>\n

Business leaders often find it disappointing to learn that, a machine learning model, after consuming precious time and dollars of investment still needs humans for routine maintenance. Let\u2019s see why this is the reality today and how companies can plan for it.<\/p>\n

A Handyman\u2019s relevance in an AI\u00a0World<\/h4>\n
<\/figure>\n

Photo by\u00a0Todd Quackenbush<\/a>\u00a0on\u00a0Unsplash<\/a><\/p>\n

To understand why models still need humans, we\u2019ll first start with a simple English explanation of how a machine learning model works, on the inside.<\/p>\n

We\u2019ll then cover 4 key scenarios where models usually scream for human help, based on learnings from the work we\u2019ve done at\u00a0Gramener<\/a>.<\/p>\n

Finally, we\u2019ll conclude by listing specific steps that businesses must plan for, to fail-proof their implementations and keep their models sane and successful.<\/p>\n

A sneak \u2014 peek into how typical ML models\u00a0work<\/h4>\n

Let us say you are a Telecom company looking to solve the big problem of customer churn. You\u2019d like early\u00a0predictive<\/em>\u00a0warnings on those customers who would migrate out in the coming month.<\/p>\n

You duly collect a dozen data feeds about customer demographics, purchases, subscription plans, and service interactions. You hand over these 100+ attributes from the past few years, which run into millions of data points.<\/p>\n

Data scientists spend their days and nights analyzing and making sense of all this data. They then build and engineer models that can predict, say 8 of 10 customers who would eventually leave. When piloted, this works beautifully and lets you focus on the task of retaining these customers. All is well, so far.<\/p>\n

Let\u2019s pause now, and unwrap this magical model to inspect its internals.<\/p>\n

<\/figure>\n

Photo by\u00a0Pierrick VAN-TROOST\u00a0on\u00a0Unsplash<\/a><\/p>\n

From the 100+ factors of customer attributes that you supplied, you\u2019d find just the key 3 to 5 used by the model. For example, these could be, \u2018contract tenure\u2019<\/em>, \u2018whether movies were streamed\u2019<\/em>, \u2018kind of service complaints reported\u2019<\/em>, or more non-intuitive factors like \u2018payment mode\u2019,\u00a0<\/em>or the \u2018number of dependents\u2019<\/em>.<\/p>\n

After studying the strength of all signals, algorithms usually end up using a tiny set of factors (or maybe just 1!) that are most related to customer churn.<\/p>\n

The true essence of all classification models can be encapsulated as:\u00a0\u201cJust give me a, b, c and I\u2019ll tell you the chances for event z to\u00a0occur.\u201d<\/strong><\/p><\/blockquote>\n

That\u2019s all a model needs, and it does just that, at the unit level.<\/p>\n

No, it doesn\u2019t always crunch a million data points. No, it doesn\u2019t model the customer brain nor does it perform any eerie reading of the human psyche. And deep learning models are not an exception either.<\/p>\n

The entire discipline of machine learning is about identifying those few factors (predictors<\/em>) and then figuring out their relationship (imagine a formula<\/em>) to the outcome (target<\/em>).<\/p>\n

<\/figure>\n

Models \u2014 Expectations vs\u00a0Reality<\/p>\n

4 Scenarios that shake the fundamentals of a\u00a0model<\/h4>\n

Now, you get a sense of how fragile the internals of a model can be, in spite of the apparently sophisticated facade. Unfortunately, people often assume that machines have a grounded, comprehensive understanding of the situation.<\/p>\n

To be fair, we humans also base complex decisions on a seemingly small set of factors. However the brain has a superior ability to play with these factors, or their relative importance for a decision. Models have ample catching up to do.<\/p>\n

Here are the key scenarios that call for models to get back into the classroom:<\/p>\n

1. Change in data structures<\/strong><\/h4>\n

Every organization\u2019s technology landscape is continuously evolving, with application upgrades, new tool purchases, and process changes. These impact the data formats and pipelines feeding the model, needing periodic updates.<\/p>\n

2. Availability of new intelligence<\/strong><\/h4>\n

Businesses routinely discover new feeds to tap into, such as the customer\u2019s online behavior, social feeds or other contextual intelligence. The model must be re-evaluated to check if these new inputs could be stronger predictors.<\/p>\n

3. Change in business\u00a0dynamics<\/strong><\/h4>\n

With a change in business strategies, customers could now be drawn to your company for its new rich product features, and no longer for the price economics. Shouldn\u2019t the model be taught this new dynamic to stay relevant?<\/p>\n

4. Ongoing refresh and validation<\/strong><\/h4>\n

Assuming none of the above changes, the model still needs a health check, say every few months. The model must be re-trained with fresh data to avoid going stale, and accuracy may dip over time calling for some technical tweaks.<\/p>\n

<\/figure>\n

Business Scenarios that call for a retooling of Machine Learning models in production<\/p>\n

What does it take to keep the models\u00a0smart?<\/h4>\n

Having established the need for support, here are the ways that companies can plan to keep their data science investments a.k.a machine learning models refreshed and operational:<\/p>\n