Artificial intelligence (AI) is here, and it’s growing — fast. Accenture estimates that by 2035, AI could boost average profitability rates by 38 percent and lead to an economic increase of $14 Trillion.
To cling on to the coattails of this enigmatic technology, brands are clambering to claim that their products contain AI. Sometimes the claim is justified, but other times, it isn’t.
What is the Difference Between AI and Algorithms?
An algorithm is a set of instructions — a preset, rigid, coded recipe that gets executed when it encounters a trigger. AI on the other hand — which is an extremely broad term covering a myriad of AI specializations and subsets — is a group of algorithms that can modify its algorithms and create new algorithms in response to learned inputs and data as opposed to relying solely on the inputs it was designed to recognize as triggers. This ability to change, adapt and grow based on new data, is described as “intelligence.”
“AI at maturity is like a gear system with three interlocking wheels: data processing, machine learning and business action. It operates in an automated mode without any human intervention. Data is created, transformed and moved without data engineers. Business actions or decisions are implemented without any operators or agents. The system learns continuously from the accumulating data and business actions and outcomes get better and better with time,” said Niranjan Krishnan, head of data science at Tiger Analytics.
Dr. Mir Emad Mousavi, founder and CEO of QuiGig, further explained the difference between AI and algorithms. According to Mousavi, we should think of the relationship between Algorithm and AI as the relationship between “cars and flying cars.” “The key difference, is that an algorithm defines the process through which a decision is made, and AI uses training data to make such a decision. For example, you can collect data from thousands of driving hours by various drivers and train AI about how to drive a car. Or you can just code it [to say] when [it] identifies an obstacle on the road it pushes the break, [or] when it sees a speed sign, [it] complies. So with an algorithm, you are [setting] the criteria for actions,” he explained.
On the other hand, Mousavi said that with AI you, “would not tell the computer what to do because AI determines [what action to take based on the] data that says this is what people almost always do.”
Related Article: 7 Ways Artificial Intelligence is Reinventing Human Resources
What are the Pros and Cons of AI?
Mousavi highlighted how AI can help to streamline many processes. “AI can make life easy by automating actions and making processes more efficient, even learn things from our day to day that we don't necessarily notice. AI can [also] scan tons of data and use that as a basis to quickly make decisions for any new situation based on that history of patterns.”
Another advantage, as pointed out by Grant Ingersoll, CTO and co-founder of Lucidworks, is that AI technologies can “adapt” to previously unseen data and make decisions “without requiring new code to be written.”
However, Ingersoll noted that running AI technologies tend to be resource-heavy. “The main con to AI approaches is that they often require a lot of data and upfront compute power (for training models) to get started, or at least previously categorized data that can be expensive and cumbersome to obtain. And, even though AI helps to make life easier, Krishnan has stated that one of the “pitfalls” of AI is that it uses a lot of “black box” techniques. “These methods have an innate tendency to perpetuate and amplify biases present in the data with respect to factors such as race, gender and education,” Krishnan said.
As it turns out, AI is also known for adopting unsavory behaviors, failing to discern political, social, and at times, even objective correctness from incorrectness. “AI invariably places women, African-Americans, and other racial minorities at a disadvantage when it comes to consumer finance products like credit cards, loans or insurance. Left to its own devices, AI could lead to consequences that include business actions going against corporate values of the company, damage to the brand, breach of compliance requirements and expensive legal violations,” Krishnan stated.
Mousavi also highlighted another drawback, explaining that “AI [currently] has no ability to think out of the box. It only acts based on prior data and would not have an answer [to] new unique circumstances.”
Mousavi gave the example of an AI-enabled, self-driving car that would choose to protect the life of the passengers over the lives of pedestrians. “if you are driving a vehicle and realize [based on] your speed you will hit a group of kids and possibly kill them, you may decide to risk your life and hit the side guardrails on a highway. However, an AI-enabled self-driving vehicle that is programmed to protect its passengers will never do that,” Mousavi explained.
Related Article: How Artificial Intelligence Will Impact the Future of Work
What are the Pros and Cons of Algorithms?
Krishnan shared how algorithms offer more control and transparency in comparison to their AI counterpart. “Traditional algorithms range from simple business rules to highly complex decision engines that require greater involvement of data scientists in tuning, maintenance and re-calibration. As a result, they afford greater transparency and control than AI that runs in auto-pilot mode.”
However, with more control comes a higher degree of responsibility, as Mousavi explained. “The programmer must include all the rules and regulations for the algorithm to work properly because it has no common sense and no idea of things that are obviously wrong to us because the program does not understand it. You won't need data to develop an algorithm, but the algorithm needs to be perfect with very specific and clear action plans to properly work.”
What Are the Best Use Cases For Each One?
Krishnan explained that taking the AI route will work best if it meets the following criteria:
- Cost of slow decisions is high (i.e. decision-making scenarios where speed is critical)
- Cost of wrong decisions is low
- Data size is too big for manual analysis or traditional algorithms
- Prediction accuracy is more important than explainability
- Regulatory requirements are less
“Use cases involving product recommendation, targeted advertising, marketing campaigns, customer servicing and predictive maintenance are well-suited for AI applications,” Krishnan said.
As for algorithms, Krishnan advised the following criteria:
- Cost of slow decisions is not high
- Cost of wrong decisions is high
- Data size is small, or at least not too big
- Explainability is critical
- Industry environment is highly regulated
“Use cases in insurance underwriting, claims processing and credit risk heavily favor the use of algorithms that are highly controlled by data and decision scientists,” said Krishnan.
What’s your take on the AI vs. algorithm debate?