|
|
This edition of TMAI Premium is free for Standard Edition subscribers.Â
It's been five weeks without a free edition! We got caught up in an exciting set of discussions in Premiumlandia. A crucial reason your analytics is not effective is that often data acts as a depressant. In #354, I'd shared a set of very specific strategies to avoid your reports and dashboards becoming a bummer. In #355, The Baklava of Brand Measurement, my recommendations help you fix the alignment problem between Purpose and brand lift studies, campaign lift studies, and brand trackers. Building out Brand Measurement's Ladder of Awesomeness was such a delight - a $10,000 ROI for your $100 Premium subscription. :) | | TMAI #358: AI Revolution: An Activation Framework. -P1
| Suddenly, it feels like Artificial Intelligence (AI) is everywhere!
Anyone with a .ai domain is getting tens of millions of dollars in funding. From street side hawkers in Singapore (they are amazing, by the way!) to train stations to Media agencies… EVERYONE has an “AI Strategy.”
Overnight, anyone with access to SenseTime or DallE or a newspaper is an AI Expert.
CEOs are not immune from the fever, they are demanding their AI Expert immediately “AI” logistics, “AI” ad delivery, “AI” movie-making, “AI” luggage, “AI” higher bonding adhesive, and, of course, “AI” tea. | The Downside of the Upside.
When revolutions arrive in tech, as they do every 20ish years, the hype results in:
A. numerous projects get green lit that ultimately fail (sucking people and money dry at a huge opportunity cost),
B. some projects yield marginal value which fails to overcome the old process/product, and
C. one possibly delivers an advantage that survives six months – sadly as the company/CEO is looking at A and B, they fall short of recognizing / tripling down on C.
Having lived through 2.5 tech revolutions, as I type this sitting on the Elizabeth Line train taking me home in London, everything above is happening in your company right now.
There is a bigger danger this time around.
With AI new possibilities are emerging at a warp speed over the last year, along with an explosion of FUD,  and “AI Experts” claiming GLORY (!) or BUST (!!). That triple combination exponentially increases the chances you are going to end up with A and B outcomes above.
And. It is not just you. Look at Google, and be worried. It had one of the most expansive, sophisticated, bleeding-edge AI strategy – executed by the very best minds in the field/s, who literally invented foundational tech like Transformers which is responsible for the warp speed acceleration of the last nineish months. Yet. Outside looking in, it has felt like Google has been caught with its proverbial pants down re an effective AI strategy.
As someone who spent so much time at G, who had the opportunity to build ML stuff, who worked with folks using ML to deliver intelligence, who was genuinely wowed by Demis' work building AGI (!)… I know first-hand the pants down part is unreflective of reality. Google, still has the best at its disposal: People, Machines, Quantum, Algorithms.
Having the best of everything did not prevent a short-term strategic miss by the collection of Leaders at Google. | An Intelligent Start to Intelligence.
Imagine the chances of your company getting this right, with close-to-none of the advantages of G.
For the rest of us mere mortals, it is crucial to take an intelligent (strategic) approach to intelligence. By applying care and thoughtfulness, you can massively increase the chances that you get max ROI from any investment you make in intelligence.
From my experience – success and failures – I’ve identified four specific lessons to prevent A, B, C outcomes from above:
1. No matter your level at your company (BUT especially if you are VP+), you NEED to have first-hand knowledge along with some first-hand experience. Â Both have never been easier.
If you don’t know it, you can’t imagine the future.
2. Be extraordinarily careful you listen to. It can be hard, tune your radar to be able to detect AI Experts from those who pay AI Experts on TV.
Reflect carefully if you should even listen to me – literally listen. [If you are a Premium member, refer to AI/ML TMAIs #52, #58, #100, #315, #335, #336 to assess.]
3. [If you guessed I would recommend this, you win a prize!] Have a framework you can use to focus your attention, to imagine the possibilities, and reject distracting AI Stuff bubbling up inside your company.
4. To earn my eternal adoration: Before you greenlight as single Revolutionary AI Initiative, do your best to quantify the upside. With AI, this is not always possible for every company. It is certainly not possible for every AI project. You will have to take some risk. But, the mere act of trying to quantify will lead to a productive discussion.
​​​​​
Two examples:
X. Currently, our costs for providing service Tx is $30k/Mo, using AI to do part automation will cost $100k, but save $7k each month.
Y. We don’t know what the outcome is, this feels super-hot right now, we are going to marshal 18 people, we will try a project in area M, area N, area O, and area P, we know the people and money investment required, $3.9 million, but we won’t know until the end what will come of it.
In each case, maybe you greenlight it or possibly you kill it. It depends on your company, leaders, and honesty levels in both. But. Hype won’t dictate the outcome, a little bit of data and a lot of conversation will. That’s a good thing.
No matter how confident you are, please appreciate that almost everyone kicking off Revolutionary AI Initiatives at companies around the world is going to waste almost all of the people, money, and opportunity. If you want to be the outlier, do 1, 2, 3, and, definitely, 4.
This series of newsletters is focused on #3 – the secret ingredient that will have you make smarter choices, make them in the right order, and ensure AI actually becomes a competitive advantage. A framework to ground us first, and the series continues in Premium editions over the next two weeks. | | The What-Type-of-AI Framework.
It can be extremely clarifying to look at the (exploding) world of AI possibilities through the lens of: What mode are we in when using AI?
Is it our friend or enemy?
Kidding.
Friend. I think. :)
As I reflect on the possibilities today, and I reserve the right to change my mind as new possibilities arise, I see three modes: Tool. Co-Pilot. Muse. |
|
The category of AI possibilities contains countless Tools. Algorithms which can chew through an entire problem/opportunity space, the human needs to just wait at the other end for the answer/decision/output and, maybe, work on activation (though often even this can be automated).
A complex example: Protein Folding. Our industry friends at DeepMind and Facebook have a tool, it can predict structures of hundreds of millions of proteins, scientists need to wait at the end for the output, discover how we can all live forever (or in the short-term accelerate cures for heartbreaking diseases). Â A simple example: Using algorithms to find unknown unknown patterns hidden inside five years of failures, and success, in your brand marketing data. Humans just wait at the other end to discover giant mistakes and oversights.
The second category is AI as a Co-Pilot. Present alongside you, all the time, helping you do what you are doing better, smarter.
A complex example: Using Co-Pilot in GitHub to suggest code and entire functions in real-time, right in your editor as you are programming. (It is brilliant and cool. If you’ve never tried it, please give it a whirl. Seems almost magical.) A simple example: Word and Gmail and all sorts of programs (Flowrite, Compose.ai) will offer to complete your sentences, fix sentence structure, or help you paraphrase.The third category is AI as a Muse. You don’t have a starting point, or you have a sketch of an idea or a wish, but you don’t have the details (or even skills to write/build/draw/create). AI to the rescue.
Midjourney acts as a muse. DALL-E inside Bing acts as a muse. Bard acts as a muse. If you would like to create something, pop into one of these tools, and they will paint a painting, write a love story involving two robots on Jupiter, suggest a trip along the silk route for 14 days under $3,000 (an actual thing for our family!), and more. It takes some effort to get out exactly what you want, like me, you are going to learn to become a great prompter, but you get back way more than you put in. (Tulip fields in Holland, Ukiyo-e style. So pretty!)
As you make your wish list of all the things you want to do with AI, it is super helpful to pragmatically evaluate the mode. AI as a Tool, Co-Pilot, or Muse?
It will help you think through the purpose, the solution space to explore, set the right expectations/scope up front, buy or build, and the resulting output/impact.
Start smart.
[You’ll note the AI as our Overlord mode is missing. It is not a worry of mine. For now.] | | AI’s Activation Framework.
With AI opportunities everywhere all at once, it is dangerous to just go at it. Or, worse, go chasing shiny objects where the impact on your business is either non-existent or poorly understood. Doing so, at the cost of what is clearly feasible today, can be career ending.
To help you identify the winning path with AI, I’ve created an activation framework. The scope is limited to my expertise in marketing, analytics, engineering, and tech.
I’ve clustered a set of related AI solutions, and, to make choosing a breeze, I’ve rank ordered each cluster based on my assessment of what’s real to needs a bit more time to cook to (#5) has been hyped a bit too much.
Opportunity cluster #0, is super important. It is the implications of AI on Life itself. It is AI as a Tool and Co-Pilot. The set of possibilities here are so wide-ranging (should love be restricted to humans?), so magnificent (imagine every human having a choice when it comes to death!), and, to some, so distracting, that I am going to acknowledge it exists and move on today. (I’ll cover all seven sub-possibilities in two weeks.)
Opportunity cluster #1 is the one with exciting immediate possibilities, you can activate them yesterday. The incredible world of generative AI, mostly AI as a Muse (for now). |
|
SEO is a wide-ranging art. Generative AI solutions like Bard, ChatGPT, Jasper.ai, Claude, etc., allow you to meet a collection of your SEO needs algorithmically and at scale.Â
Try gAI for product descriptions, landing page content, relevant blog posts and newsletters for your customers. None of these are completely trustable quite yet, but they are excellent at providing quality first pass level outputs. From there, a human – now only spending 40% of their time on all this activity – takes it the rest of the way.
I am excited about experiments, re content, for using gAI to personalize content for an individual (a holy grail that has escaped very solution for 20 years). Throw anonymous first-party cookie data from GA into a gAI, Â connect back to the site, customize future content for every individual! Early days. Loads of promise.60%+ success of any marketing campaign is driven by the creative. Building creative is expensive. Because so much of your creative fails pre-testing (you are doing creative pre-testing with Human Made Machines, right? right?), you want to build more concepts, more executions of that concept to let the imagination of your creative directors to flow freely. The problem: Money. Time. It is very expensive to play this game.
Enter my BFFs Stable Diffusion, Dream, and Midjourney (showcase – mind-blowing!). Your creative team just needs a bit of training on optimal prompting can generate as many concepts as they want, rather than discarding wireframes. They can build out as many executions of those concepts, and pick the most resonant ones for creative pre-testing. Costs go down, creativity goes up, speed increases.
Any business can only shoot pictures and videos with x number of people, sometimes at the cost of diversity (of all types – age, gender, geography, sizes). It is often just economics. AI generated synthetic humans, to alleviate this problem. The “people” you see on Lalaland don’t exist in the real world. But, in every way, they look real. Levi’s can now have 15,000 pictures of models of all types for their clothes. Here's an example, synthetic human... |
|
Extending this possibility, for a very low cost, any company big or small can produce videos with synthetic humans to create more engaging user experiences using a solution like Hour One or Synthesia. #dieboringtextdie
The possibilities are endless.
The first time I used Midjourney last year to create website designs, I was blown away by how good I was at website design! Ok, ok, ok, not me so much. Midjourney. : ) For a simple but wonderful demo, checkout this video by Monika Buchelt using ChatGPT and Midjourney to create a website. I have far fewer skills than Monika, and I did pretty ok. The ChatGPT-powered site structure recommendation was a fantastic starting point (only needed 20% tweaking by me).
If you have 30 seconds, go to Durable, click Get Started, pick a category (Restaurant, or whatever), type in a business name (Nanyang Kitchen, or whatever) and wait to be wowed. (Click Regnerate Section if you don’t like something). The end result will be far from perfect. But, remember, 30 seconds ago, you did not have a fully created website design and content! If you have a bit more time, try 10web.io.
Important: Website designers, UX professionals, are not all about to be jobless. (Not for a while.) Rather, what’s exciting to me is that it makes site design, content, video creation, way, way, way more democratic. And for Designers and Professionals, instead of making 2 options for a client, make 9.
For what is possible with fully AI created movies today, checkout JANUARY and Last Stand. Neither is going to win Oscars yet. Imagine your product demos, brand stories, certainly video ads, and so much more that could go from prompt to running on your site or Facebook’s ad networks in a few minutes.[Note: I am crazy excited about the implications of NeRF (Neural Radiance Field). From converting photos and rendering into photorealistic video (see this) or "reskinning the real world" (this is incredible), the possibilities emerging are incredible. The next Avatar movie will be $1.5 billion cheaper. :)]
In none of the six sub-possibilities with gAI are we at a stage of perfection where no humans need to be involved. Not yet. What gAI is doing, as a Tool, as a Co-Pilot, is increasing possibilities, democratizing access, cutting a massive amount of cost out of the equation, and delivering GIANT sized opportunity for process optimization and efficiency.
That is why I am So Excited.
That is what we are doing at Croud. | Up Next.
Five more clusters of exciting opportunities, identified, prioritized, for you to take advantage of.
We’ll start next week with Smart Algorithms, all the astonishing ways to transform your practice of analytics with AI.
Get ready to discover unknown unknowns!
[If you are not a Premium member, it is easy.] | | Bottom line.
Are you using AI as a Tool, Co-Pilot, or Muse?
Is the first set of your initiatives taking advantage of all that’s already possible with Generative AI?
If you are, ain’t it radically awesome?
Carpe Diem!
Avinash. | Committed to investing in your professional growth?
Upgrade to TMAI Premium here - it is published 50x / year. | |
|
|
[ Subscribe] Â |Â Â [ Web Version]Â Â |Â [ Unsubscribe]
                            ©2022 ZQ Insights  |  PO Box 10193, San Jose, CA, 95157, United States of America |
|
|
|
