Pick none: Death, Sorrow, Disillusionment
Ignore the hype; we're ok, AI magic is coming soon
It’s consulting hype and you don’t need it.
Trough of Disillusionment is from Gartner - big tech consulting firm. Not to be out-doomed, wannabee VC’s came up with The Innovation Valley of Death AKA VoD and the Trough of Sorrow AKA ToS.1
These are consulting frameworks that predict the progression of a new technology as it matures. I keep seeing AI doom articles using these frameworks as people get impatient with AI not having changed the world … yet.
Here’s Gartner’s hype cycle, guess where we’re at:
Been there; I rode the consulting train straight into the Pit of Doom™️ AKA PoD and out again2
In my consulting heyday (Plateau of Productivity AKA PoP?) I was what consulting firms call a brain-on-a-stick or just a smart guy fixing stuff.
I did some magic in my time as a tech wizard. A small business that owned a bunch of parking lots in downtown Oakland fired their network administrator. He did something bad - they didn’t tell me what. Of course he left in a huff, changed all the passwords and locked them out of their systems. No parking for you!
I had them back in the car parking business in under 30 minutes and I (my consulting company at the time) charged them a bunch because that was good magic. It wasn’t the time I spent or the graphs I drew, it was my big brain they paid for.
One more magic trick for when the world was running out of computer addresses. To be a thing on the the Internet you need a unique name. That’s why you have silly emails and account names like hopefulplatypus_481943 or some abomination of your name with periods and numbers or cryptic abbreviations.
Just like you, every computer has to have a unique name. Computer names are numbers called IP addresses and back then the world was running out of them. My client this time was The San Francisco Chronicle. Stop the presses! The Chron couldn’t manage their network anymore because of this.
This project took me and a crew all night scurrying through the Chronicle’s big old building in downtown San Francisco. We built out a new IP address scheme4 and gave every computer there a new name. In the wee hours when we all got a little loopy it got hairy with some obscure systems - that actually did drive the printing presses - not liking their new names, but we figured it out and the papers went out in the morning.
Not everyone has a big brain or a brain that cares for IP Addresses or Sysadmin back doors. If you run a small consulting shop you find the best brains in your niche and train up. When you run a big consulting org, you can’t just rely on the big brains. To level stuff out you sell Frameworks and Methodologies and Reference Architectures.
I worked in consulting for a long time at Microsoft. Initially I kept to my brain-on-a-stick but as Microsoft Consulting Services grew, I moved into building out repeatable reference architectures and offerings until the cloud came along and it became too hand-wavey for me and I moved over to build products.
Business Consulting is big business to the tune of $800B a year and what they do gets more and more hand-wavey which takes us back to Methodologies and the like.5
Consultants bring an outside perspective and expertise but they’re not for everyone. Steve Jobs didn’t like consultants. At at talk at MIT’s Sloan Business School6 he said consultants were like an imitation Banana:
I think that without owning something over an extended period of time, like a few years, where someone has a chance to take responsibility for one’s recommendations, where one has to see one’s recommendations through all action stages and accumulate some scar tissue for the mistakes and pick one’s self up off the ground and dust one’s self off, one learns a fraction of what one can.
You do get a broad cut at companies, but it’s very thin. You never get three-dimensional. You might have a lot of pictures on your wall, you can say ‘Look, I’ve worked in bananas, I’ve worked in peaches, I’ve worked in grapes.
But you never really taste it.
Dude liked his fruit.
So what’s with all the AI consulting hype? It’s still early days wandering through the Trough of Disillusionment, but here are three big AI brains making moves to noodle on:
Sam Altman runs Open AI, makers of Chat GPT and still the lead dog in the AI race. Sam is so convinced that AI will fundamentally change society, he created a research arm that just completed an 8 year study on the impact of Universal Basic Income. AKA if we just give everyone money, will the people who lose their jobs to AI be happy?
Spoiler - money doesn’t fix everything - Universal Cash Study
Zuckerbot AKA Mark Zuckerberg released their latest AI model called Llama as open source or freely usable. Zuckerberg predicts Billions of small business AI agents will come out of this. If you buy into this prediction, soon Wirepine will be able to wire up an AI agent on your website easy peasy. You wont ever talk to a customer again.
This is an odd video with Zuckerbot apparently just rolling in from suring off his Hawaii compound with racoon eyes from the sun. Bajillionaires are weird.
I talked about Marc Andreesen a few weeks ago in Hunting Unicorns as one of Tech’s Venture Capital kingmakers. Here he compares AI to the emergence of the Microprocessor. Because AI isn’t deterministic like traditional computing but rather probabilistic, it changes everything. Big statement, big implications.
Here’s the snip from the Substack podcast with Chris Best, Substack CEO:
Will the big brains get any of this right?
Here’s another way to think about it. Lets compare a 4 year old baby to the the latest AI large language model’s (LLM).
Every LLM releases with specs and with each release the specs get more impressive. For example, the number of parameters a model was trained on is important. The one Zuckerberg was talking about Llama 3.1, released this week and was trained on 405 billion parameters.
That’s a big number but by the time a baby is four they have had orders of magnitude more ‘training’ than that. A baby takes in everything it see’s, hears, tastes, feels, etc. Everything! If you’ve ever read a book to a kid, that’s packed with 100X more information than just the words. Sound, pictures, expression. All AI gets today is text. No tone, no context, no pictures. That’s why at their core today, LLMs are text prediction machines. Text is inherently limited. Upcoming models are training on video and audio.
Finally, all the hype is around generative AI because that’s the thing we can play with directly. Google just released Gemini, Microsoft has Copilot. If you’re streaming the Olympics you’ve been bombarded by commercials of these products built on generative AI. However gen AI is but a piece. There is lots of amazing AI tech coming behind the scenes that’ll just make our lives better.
Here’s a cool example:
We’re still at the beginning, ignore the hype.
best, Andrew
This is Wirepine’s second consulting methodology after the Wirepine Marketing Pyramid or WMP I introduced last year in Lego my Logo.
My Discord account
We switched the Chron over to a private address scheme behind a firewall but nowadays we’ve moved from IPv4 to IPv6 which provides a ridiculous (340 undecillion addresses) so now we good.
A few of the top shelf business consulting names you may have heard of: McKinsey, Boston Consulting Group AKA BCG and Bain & Company.