Navigating the Gen AI Hype: My Criteria for Meaningful Innovation
Balancing Excitement with Practicality, Applicability, and Impact
Windows 95, Microsoft Office, Netscape Navigator, Email, Google.com, WinAmp, iTunes/iPod, WhatsApp, Instagram, Kayak.com, Airbnb, Uber, Zoom … a non-exhaustive list of killer apps I’ve come to appreciate in my lifetime. Killer apps are software or devices that are so useful, innovative, or desirable that it drives people to adopt a particular technology, platform, or device just to use it. As we are in the early days of Generative AI, have you found a killer app that is so compelling and offers something unique that you cannot get anywhere else?
While there have been many exciting Generative AI tools/models that have come to market (e.g. ChatGPT, Gemini, Claude, Llama), primarily focused on semantic search and an expanding portfolio of music and art tools, we should actively experiment, learn, and share our observations as we seek to design and launch Gen AI’s killer app. Like the early days of Web 1.0, it will take all of us to actively gather, compare, rank, and make sense of the advanced in Gen AI to brainstorm, fund, and realize this generation’s killer app.
Setting your criteria
Novelty is the primary feeling that we feel after experiencing a new AI tool or demo. Remember the first time you tried ChatGPT, created AI art, or deep-fake? Awe inspiring. New. Fresh. Cool. But does the quality of being new, unique, or different translate into utility for you or society?
As we explore Gen AI, it’s important to set the criteria in which you will evaluate a new application, platform or tool. Defining criteria up front will help you properly analyze the available options by narrowing down your choices based on your desired outcomes. Having criteria prevents you from comparing and contrasting options, which provides relative reference, but may not align with your outcomes.
Here are my criteria for Gen AI:
Amplifies one of my strengths (something I am good at and enjoy doing)
Fills a gap in an area where I'm not strong, making me more productive or improving quality
Easily integrates into my life
Brings positive change
Privacy and security built-in
Shares logic or references in how it comes to conclusions
Conduct Your Own Turing Test
As Gen AI seeks to imitate human intelligence, we can consider conducting our own Turing test on the results and outputs of Gen AI tools. The Turing Test, proposed by Alan Turing in 1950, is a measure of a machine's ability to exhibit human-like intelligence. The test evaluates whether a machine can engage in a conversation that is indistinguishable from that of a human, effectively "passing" as human in communication.
An important consideration for Gen AI is that it’s built on an amazing wealth of information on the internet. This includes text (web pages, literary works), user-generated content (social media, user reviews), images, audio, and structured data. But this can include true and false data that is incorporated into models and results of Gen AI. Let’s take results with caution. If we fed Gen AI literature of the 14th century, it would tell us that the Earth is the center of the universe with the Sun, Moon, and stars circling it.
Another test I like to conduct on semantic search results (e.g., Gemini, ChatGPT) involves activities where I consider myself nearly an expert, allowing me to effectively evaluate the accuracy and quality of the generated AI responses. Here is a sample of my observations on detailing cars and training for a triathlon.
When mapping these results to my criteria, ChatGPT stands out by allowing me to quickly and directly apply the outputs to my tasks. I also appreciate Co-Pilot's straightforward and effective pro-tips, as well as Claude's tree-like structure for diving deeper into specific elements of the search results.
My Curiosity of Music from Pre-Web to Web 1.0 to Web 2.0 to Web 3.0 to Today
Being curious and understanding what does / doesn’t work as well as the technology behind different iterations is important for us as we experience and make decisions. I remember the first time I sat down and learned how to use our family’s record turntable. A late 70s / early 80s AKAI stereo system with an amplifier, equalizer, record player, and a cassette deck. Fast forward to today, our consumption pattern and platform for music is completely digital, on-demand, and hosts a library unimaginable a few decades ago. I’ve come to appreciate the social and technological change of music distribution and consumption, by being involved in its journey. Here are a few key moments I remember:
$0.01 Columbia House Music - Get 10 CDs for $0.01! ….. and then buy 8 at full price (+ a mark-up). This was my first foray into music collecting with my first ever CD, Green Day’s Dookie. With CDs being relatively new technology, the appeal of a massive discount was a huge appeal to boost CD player sales and a move of distribution from tape to CD.
Creative Labs - Sound Blaster + SCSI-II + CD-Writer - moving from onboard motherboard speakers that could only chirp to 8-bit, then 16-bit sound really opened up my intrigue with digital audio. In connection with the same vendor, Creative Labs, we bought our first CD-write, but had to also first upgrade the BUS between the mother board to ensure uninterrupted CD writes. This was a wonderful introduction to modifying and building computers.
RIO 32MB MP3 player - as Discman CD players were popular, WinAMP popularized MP3 formats (along with Napster, Kazaa, BearShare, etc). This is when I better understood encoding and bought one of the first MP3 players to market, the RIO. While 32MB was really limiting (up to 8 songs max?), it let me experience digital music on the go without any mechanics (e.g. Discman or walkman motors). Ultimately, my first 4GB iPod was a 10x leap, the RIO player was an early entrant and first experience for many.
iTunes Digital Rights Management (DRM) - after downloading a fair share of MP3s, it became apparent that torrents and file shares were littered with low quality copies of music or laced with viruses. To complement the Apple iPod, iTunes changed the music distribution with access to label-supported MP4s offering higher quality audio, right management, and artist royalties. But why pay for individual songs if you could …
Spotify - who disrupted the traditional CD sales and digital downloads market with a streaming model that was subscription based, on-demand, and providing an immense library. Spotidy provides a convenient and accessible way for listeners to discover and consume music.
These Are a Few of My Favorite Gen AI Things
Udio is a unique AI-driven platform that creates personalized music. It stands out with its ability to generate custom music tailored to my preferences (e.g. genre, topics). Unlike other audio platforms, Udio provides an interactive experience where I can keep tweaking and modifying the song to my liking (e.g. extend, remove sections). I see myself using it to create soundtracks for home videos or social media. My daughters and created a song about our upcoming trip to Target but having to finish our goldfish snacks before leaving. It took less than 3 minutes with a simple prompt of: “write me a song about eating goldfish before going to target. Make it a Pop song.” Take a listen to, “Aisles of Smiles.” Or an impromptu song about a 40th birthday party in Las Vegas.
I also recently came across Meta’s, “Reimagine me” feature in its Instagram and WhatsApp chat. After taking 3 profile photos, you can ask Meta to reshape yourself into nearly any identity. What a great way to create content that truly reflects me or who I may aspire to be. Here are a sample of myself as a baseball play, olympian, cowboy, and more!
Now it’s time to map these Gen AI tools to my criteria. Upon evaluating the restuls, it appears Udio and Meta Reimagine primarily serve as novelties for me, while semantic search shows some promising qualities.
So What Am I Looking For in Gen AI?
I am genuinely excited about the possibilities that Gen AI holds for the future. The advancements in this space have the potential to revolutionize the way we interact with technology, creating more intuitive, personalized, and efficient experiences. However, in the midst of this excitement, I recognize the importance of carefully evaluating the applications, tools, and features that emerge. This note outlines my approach to assessing these innovations, ensuring that I don't fall prey to the inevitable hype cycles that often accompany new and groundbreaking products.
My focus is on identifying what is durable and impactful, both for my own life and, ideally, for the lives of many others. While novelty can be intriguing, I am particularly interested in solutions that offer long-term value, addressing real needs and providing meaningful improvements. By maintaining a critical perspective, I hope to distinguish between fleeting trends and truly transformative developments in the world of Gen AI.
I’ll wrap-up with my wish list:
A tool that can seamlessly stitch my vacation videos together to create an interesting movie that I’d enjoy watching with my family repeatedly. It would be great if it could be tagged to feel like a mystery movie, adventure, or documentary.
Semantic search that blends seamlessly into the applications I use today. I’d like it to be ready when I prompt it but also capable of prompting me with suggestions.
A personal coach that helps me keep track of my personal, career, and family goals. It should suggest options, keep me accountable, and especially help with long-term goals that I tend to forget or stray from.
What do you think is the unwritten killer Gen AI app? What’s the feature set you’re looking for? What’s the criteria upon which you will evaluate Gen AI?
… how will you help build the next killer app?
Great post! I'd love to have an AI agent acting as an intermediary for calendar planning between multiple work calendars and personal calendars to plan personal and work events.