Artificial Intelligence

Why Does the Hype Around AI Exceed Implementation


“By far the greatest danger of Artificial Intelligence is that people conclude too early that they understand it.”

– Eliezer Yudkowsky

The global AI market size totaled USD 51.08 billion in 2020 and is projected to reach USD 641.30 billion by 2028. Artificial intelligence is everywhere. Some businesses are using it, some are planning to, and others have an opinion.

Consequently, we can sometimes be misguided by the misconceptions and fascinations of AI resulting from its image created by science fiction and pop culture.

With advancement at the current rate, there’s little doubt that AI and data analytics can reach great potential. However, the optimists among us may miss the many challenges lined up, while the pessimists don’t even touch the technology for fear of failure.

Before assessing why AI hype exceeds its implementation, let’s look at the hype.

The Reality of AI Hype

According to a 2021 study by KPMG, three-quarters of surveyed executives agree that AI is more hype and less reality. Half of the respondents also agreed that AI is moving too fast in their industry and wished to move faster.

The buildup of AI leads companies to get into the game with a false perception of what AI can help them achieve. Without a clear understanding of what technology can and can’t accomplish today, there is a lot of risk in getting involved.

Another significant fact is that 92% of said respondents are implementing the technology and believe in its ability to deliver value in terms of high efficiency. While there is confidence in the technology, hype understates the challenges involved.

The overpromise of AI led to the coining of the term ‘AI Winter’ to describe the period of boom and bust in the 90s when the technology faded for a while. As it did, people left training programs stranded. So today, the huge market demand for AI doesn’t match the availability of proficient professionals.

Consequently, the advice and implementation a lot of companies receive are experimentative.

Why Do AI Projects Fail?

87% of data science projects never go into the production stage. And 70% of companies report minimal or zero impact from AI.

Mixing Marketing with Implementation

Often, AI vendors will engage in overpromises in their marketing, giving their clients a false sense of expectations. Companies often engage with technology vendors based on marketing alone and forget to look closely at previous implementations and results of the same sort.

Usually, a vendor may also convince a business that their AI implementation does not need a particular component when it does- just that the vendor doesn’t provide it, and the company doesn’t know.

Unclear Objectives

Artificial intelligence can be a powerful strategy once you know what you expect from it. Having a well-defined business problem with business goals can help achieve success. 

Besides, measuring outcomes from an AI implementation can be tricky as it involves building and training an AI model and experimenting with long-term trial-and-error before seeing results. A well-defined business goal can determine whether the direction of implementation is correct.

High Expectations

Business owners can embark on an AI project with high expectations. Self-driving cars, autonomous drones, and facial recognition systems are significant use cases of AI. Still, startup owner needs to look at the core business value they’re trying to achieve with AI instead of the new shiny objects.

High expectations around what AI can do for you often lead to disappointment when business owners conveniently underestimate the challenges and misinterpret the reality of AI.

Lack of Access to Talent

There are opportunities galore, and not enough experts in the AI industry who can steer the ship and take AI projects to the finish line. Hiring for an AI team can mean huge investments, and working with a vendor needs a careful vetting process.

The lack of easy access to talent (and a lot of marketing) makes it hard for companies to distinguish genuine professionals from others. Getting into it with the wrong people can mean huge irreversible losses that may have been better off invested elsewhere. 

Steps to take for a successful AI project

1. Strategize

The implementation part of AI comes later on the roadmap. First, you need a strategy that has been thought out internally after looking at various options and alternatives. This way, you are not dependent on a technology vendor to build your plan. Instead, you can approach them and directly talk about implementation.

2. Start Small

Instead of looking at AI as a business opportunity, companies can do better by leveraging it to improve business outcomes first. Begin by analyzing decision points and asking how improving this decision by 1% impacts the bottom line.

Starting small can set you up for success with AI by giving you insight into how it works. After a tiny victory, you can incrementally move on to more significant objectives.

3. Plan for Organizational Change

Artificial intelligence projects can change how business happens in your company. Are you and your team ready for it? Spend time getting executive buy-in and getting all employees on board. Explain why this project would mean a lot to your company.

A big part of implementing AI is dealing with the resistance from within. It’s natural for us to feel a little threatened by AI if we don’t believe that AI is here to augment humans, not replace them.

We created an in-depth report for startups exploring AI as a realistic possibility beyond the hype. Check at the end.

Is AI suitable for your startup?

As a startup, you spend a lot of time and effort trying to sell to potential investors, customers, or both. It’s a good idea to reflect on and consider what you’re selling to yourself regularly. Consider the following AI use cases and how they correspond to the current realities of your business.

1. Demonstrable Capability: Good

In this case, AI capabilities are built into your current product and meet a genuine business need. You can discuss the models in use and the current training and inference infrastructure and share supporting artifacts such as Jupyter notebooks (or equivalent) used to validate these models.

Alternatively, you can use commercial offerings from AWS (Amazon Web Services), Azure Cognitive Services, and others to demonstrate their integration. You clearly understand the applicability and limitations of these models.

2. Aspirational Capability: Good

Early exploration to validate the approach and prove the use case may already be underway in some cases. In any case, you may be looking for funding to build this area, which is fantastic!

An accelerator, such as LogicBoost Labs, can provide resources to move AI use from aspiration to reality at the minimum viable product (MVP) stage.

3. Redefining AI/ML: Bad

In this case, a company attempts to reframe a specific technology or technique as AI when it is not. The average person might believe that, but those in the know will see it as deceptive and impolite. It is not worth jeopardizing investment or new customer opportunities by promoting AI capabilities that do not exist. 

Furthermore, whether AI or not, be proud of your knowledge, product, and capabilities, and don’t oversell yourself. Those in the know will appreciate that.

4. Handwaving AI: Ugly

Handwaving AI occurs when a company fails to understand what current AI systems can and cannot do. Frequently, these companies don’t know how to apply AI to their product or struggle to force a connection – a situation known as a solution in search of a problem.

In this scenario, large parts of a proposed solution far too often are ultimately disconnected from any current AI capabilities.

So, which category do you fall under, and why?

Checklist for your startup:
Aspirational or Actual Use

  • What business problem is it solving?
  • Can you demonstrate the capabilities and the systems/artifacts that drive them?
  • How is the data gathered and prepared?
  • What models are you using?
  • How is the model trained, validated, deployed, and retrained?
  • Are there future scalability issues to be aware of?
  • What are the economics around model training and inference? Is this factored into your pricing model?
  • Do you have ethical and bias controls in place?
  • If used in a regulated industry or mission-critical setting, what measures are in place to ensure compliance and understand failure modes?
  • Who on the team is the resident AI expert? Do they have the skills and experience needed to evolve and refine the offering?

This checklist will assist you in determining whether AI for your startup is aspirational or in use. Describe any prototypes or experiments that are in the works. Most MVP-level product offerings are rough around the edges, particularly in AI management and deployment (also known as MLOps), and that’s perfectly fine.

It could be one of the reasons you’re looking for more information. However, no one expects a startup to have a fully automated MLOps workflow on the first day. These capabilities can be properly integrated over time.


We live in an era in which the accessibility and capabilities of machine learning (ML) and artificial intelligence (AI) technologies are expanding exponentially. It’s pretty thrilling. (For the sake of brevity, we’ll refer to this entire range of technologies as AI.)

While there is a lot of hype surrounding this topic, it is difficult to deny that AI will fundamentally change our daily lives. Startups frequently set the standard for introducing and innovating with new technologies, so no wonder that their use of AI is gradually becoming the norm.

However, as with any game-changing technology, introducing AI into the mix without first understanding can lead to unrealistic expectations, misinterpretation, and confusion.

Are you prepared to rebuild your industry? If yes, consult our AI development experts who can help your company reach the pinnacle of success through intelligent digital solutions.

Subscribe to our Newsletter
Stay current with our latest insights