Should you really stop doing AI pilots?🧪
Remember Cryptokitties? These digital cats sold for over $300,000 a few years ago. In case you missed it, worry not. This week a jpeg sold for $69 million! You can download it for free here. The premium is for ‘owning’ rights to these digital assets.
It’s NFTs or Non-Fungible Tokens that power these digital cats, digital art, and basketball video clips. Would you rather download the image or build your own YouTube playlist? I won’t blame you😀
This newsletter will take you about 4 minutes to read.
I. Spotlight: Should you really stop doing AI POCs?🧪
A recent article from Gartner recommends limiting the number of POCs (Proof of Concept) you do in order to be successful with AI. The report states that organizations now exploring AI in their operations do about 20% fewer POCs.
This isn’t entirely correct.
(Picture by Science in HD on Unsplash)
The recommendation is based on the assumption that organizations now have a better understanding of AI, its capabilities, and limitations as compared to earlier.
AI Pilots aren’t passé
Despite the industry’s rising maturity of AI, I think AI pilots are very much relevant today. Here are four key reasons why you must continue investing in them:
To check if the AI will be effective on your data: AI is as good as the data you feed it. There are great examples of AI, but will they work with your data?
To get a sneak preview of how your users will receive it: Pilots can quickly demonstrate business value to your users and get crucial early feedback.
To validate your costs before doing the full build-out: AI needs labeled data, heavy computing, and subscription costs. Pilots will help you size up your costs.
To secure your sponsor’s buy-in and budget: Organizations struggle to secure investment for AI. Pilots help demonstrate value and secure your funding.
While there is a clear need for AI pilots, most organizations get them wrong. This leads to lost time, efforts, and ultimately doubts on the very relevance of pilots.
At Gramener AI Labs, we’ve been executing pilots for the past few years. We prioritize the top client challenges in the industry. Then, we scout for the latest AI fundamental research to see which of those can be applied to tackle real-world issues.
For example, we predicted the quality of life from satellite images by applying computer vision algorithms. It was built upon the research done at Stanford. We presented this at O’Reilly AI Conference 2019 in London.
Subsequently, we scaled it into a commercial solution to fight mosquito-borne diseases like dengue that kill millions of people every year. This solution recently won the 2021 Edison Award for Innovation.
(Picture: Defeating Dengue with Geospatial AI)
How can you get the best value from your AI pilots?
We’ve learned that you need a deliberate approach to POC selection and deployment to achieve the right benefits, based on our experience, . This is one aspect where I do agree with the Gartner article quoted above.
Here are five recommendations for your next AI pilot:
Pick your pilots just like projects: Be clear on the purpose and evaluate business alignment, impact, and urgency.
Define success criteria upfront: Ensure that there is agreement on what constitutes success and how it can be taken to the next stage.
Build and iterate rapidly: Staff for cross-functional skills and run the pilot through agile sprints across a few weeks.
Ensure business ownership: Secure business involvement and don’t treat pilots as pure technology experiments.
Start with a plan to scale: Think through how the pilots will help you build a scalable production solution.
Remember, you don’t always have to throw away your POCs. Well-designed pilots can act as MVPs that you can scale up.
II. Industry Roundup:
1. Gartner Magic Quadrant 2021
Gartner published its annual Magic Quadrant (MQ) for Data Science and Machine Learning Platforms. It provides a good overview of the key commercial platforms and top industry trends. This time most players fall into the ‘Leaders’ and ‘Visionaries’ quadrants, which Gartner calls a “glut of compelling innovations” in the space. Check out this review of the Magic Quadrant or the full Gartner report (needs access).
-> Check out VentureBeat’s Review
2. Healthcare NLP Summit
This is a free conference at the intersection of Healthcare/Pharma/Biotech and Natural language processing. It features exciting speakers and has a combination of research and industry application talks.
-> Register for the free Event (April 6 & 7)
3. Is your Analytics team an Order Taker or a Decision Maker?
9 minutes | California Management Review | Nikhil Sikka, Seth Borin
To improve adoption of analytics, you must take a decision-backward approach with data. This article advocates a revamp of the traditional analytics-driven workflow. In order to place decisions at the center, here are the 5 steps you must follow. Check out this article that uses a real-world example to drive home the point.
III. From my Desk:
1. Nasscom D&I Summit Panel: Technology for Inclusion
I was part of a panel discussion on ‘Technology as an Enabler for Inclusion’ at the Nasscom Summit, this week. We discussed how technology can help promote inclusion and we talked through industry examples. The video will be out soon.
2. Coming Soon: Data Science Whiteboard Season - 2
I took up a challenge to explain the business of data science in 5-minute whiteboards. It was a big success and the 12 videos generated great conversations and were viewed 35,000+ times just on LinkedIn. Excited to share that Season-2 is launching next week!
Do you love playing computer games like World of Warcraft, Civilization, or Age of Empires? This might be a sign of great leadership, says this research paper!
Thank you for subscribing and reading the newsletter. I appreciate your attention,
Ganes.
PS: Did someone forward this to you? You can subscribe here.
My Website | LinkedIn | Twitter | YouTube