No ROI in AI?

No activity in business escapes the scrutiny of ROI, especially Gen AI.?

Global tech leaders have invested over $1Trillion dollars into Gen AI, but the results paint a sober picture. 75% of projects are never completed and over 42% of completed projects have little to show for it.?

So, where have we gone wrong, and how can we realign our expectations and strategies to realize the ROI from AI initiatives?

?

Why do Gen AI projects fail??

According to Gartner, over 85% of AI projects fail for various reasons.?

The reasons for these failures are multifaceted:

1.Lack of Clear Objectives:?

This is something I see on a regular basis. Leaders who seek problems to solve using technology seldom find success. It’s important to gain a clear understanding of goals and success metrics before using Gen AI.?

2. Data Quality Issues:?

Your AI is as good as the data. If you feed an AI system bad data, you're going to get bad results.(Garbage In, Garbage Out (GIGO). It is surprising how poor data quality can derail even the most promising projects.

3. Infrastructure Issues:

Even when pilots do go well, companies find solutions are hard to scale due to lack of forward thinking. Gen AI projects need good investment in robust infrastructure - either on your end or your vendor’s.??

4. Hyped Expectations:?

Starting Gen AI projects with unrealistic expectations can lead to disappointment even when Gen AI projects aren't delivering “groundbreaking” results.?

So, when you’re starting a Gen AI project, be aware of these major reasons why Gen AI projects fail. While all the other reasons like lack of objectives, data quality and infrastructure requirements are tactical, setting expectations is a strategic move.??

So, let’s see how to re-frame your Gen AI ROI expectations??

I try to classify Gen AI projects into 3 types.?

So, if I were to start Gen AI project again, here’s what I’d do:?

• Quality Data is the only way: Before even doing anything, I’ll do a data quality audit. I’ll take a look at the enterprise data to gauge its useability for Gen AI projects.?
• Start Small but Scale Smart: I’ll first begin my pilot projects only for quick wins - with scalable infrastructure. Once I build organizational buy-in, I’d scale these quick wins and then go for the other kinds of use cases.?
• Upskill Your Workforce: Once I’ve scaled one quick win, I’ll invest in AI training to improve Gen AI project adoption rates in the company.?
• Iterate till you succeed: I’ll make sure to keep experimenting with different use cases to see which nes perform well before scaling them up.
• Measure Beyond Financial Metrics: Along with financial improvements to measure ROI, I’ll also measure intangible benefits like improved customer satisfaction, employee productivity, and innovation potential to gain buy-in.?

Remember, the journey towards AI ROI is a marathon, not a sprint. Set realistic goals, expectations and choose the right use case to get the most out of AI.?

During my courses at CMU and UCI, I created a ML4B: Machine Learning for Business Canvas drive AI projects to deliver successful ROI. Several of my former students swear by it. I am making it available for free here for a limited time.?

P.S. What has been your biggest challenge while proving AI ROI??

I’ve started this newsletter to answer the most common queries I’ve received while working with tech leaders like CIOs, CPOs and CDOs over time while implementing Gen AI projects.

I’ll cover topics like security concerns, ROI traps, vendor screening and more.

If you’d like me to cover a specific topic, let me know in the comments below!?