Publish Date
Nov 1, 2024 08:00
Word Count
1,205 words
Status
Published
Campaign
Excerpt
Canonical URL
Do not index
Do not index
Notes
Know Your Limitations
The limitations of any tool are as important as its features. You can’t use it effectively until you know both. And like styles of fine art and genres of music, I find that the limitations of software inspire human creativity.
Granted, today’s limitation can be tomorrow’s feature. But remember I said how the big players can afford to change the game more than once? That’s how this plays out: in waves. With each wave, all the pieces are moved again.
So here are some observations on how I think the current wave will transform my career. I hope you find some inspiration for your own game plan.
1. It needs CRM
In fact, organisational AI needs much more than CRM. There’s a Salesforce product called Data Cloud which gives you some idea of how much data is needed for this wave to reach its potential. CRM is a foundational step however.
Although I knew this, I didn’t get it. This dependency by AI will have unforeseen consequences on the CRM industry. It means that every CRM customer should P0 their data management, and it means that all businesses need a CRM, even if they’re too small to have reached that conclusion independently yet.
Good for CRM consultants? In many ways, but also it’s breaking the silo of expertise. There’s a big gap opening in the bottom of the market, and ultimately small businesses will need to learn the basics by themselves.
But here’s the thing: that doesn’t just transform my career; it transforms yours. To get the most out my personal AI, I had to build my personal CRM. We need to talk about the personal benefits.
2. Its Memory is Much Smaller Than its Knowledge
Thanks to sci-fi, I had a preconceived notion of an AI personality that persisted from one session to the next. Instead fleeting, isolated chats with multiple agents requires a new way of working. In curious ways, it’s highly dependent on human memory, human consciousness and human decision-making.
This factor dictates the nature our work with Gen AI significantly. I discussed this in my previous post: the context of a real-time chat is more akin to ‘collaboration’ than to ‘delegation’.
Granted, fundamentally it’s a hardware limitation, and obviously we expect memory capacity to increase for AI, as we’ve seen it increase for other hardware. However, AI hardware is particularly expensive at this stage, and that market’s not competitive as yet. Therefore, I’d expect this distinction between “memory” and “knowledge” (my terms) to remain significant for a while - even if it becomes more of a pricing factor as longer memories come on the market alongside the present wave.
How’s this significant to our careers? I don’t know exactly. It affects the nature of human-AI collaboration, placing a high value on our ability to hold a train of thought for a whole chat - and from one chat to the next.
Here’s a thought as to how employees and jobseekers could play that one strategically:
- Step 1, it places a new ‘stewardship’ responsibility on us, as a consistent presence across multiple generations of agents. This requires our expertise.
- Think: I couldn’t do your job, even though we both have access to ChatGPT. What’s the difference between us?
- Step 2, it creates new work for us, as we create new frameworks for meeting this responsibility consistently. (I’m going to call that ‘Agent Onboarding’; I suppose I’m not the first.)
- Think: Your boss knows only one prompt for each of your jobs; you know a hundred.
- Step 3, it creates new work for us again, both in iterating on the frameworks and (unpredictably) from the new paradigm that emerges.
- Think: What’s that new process you wanted to implement, but you weren’t able to share your vision?
- Think: What does Super You need to do, if only you didn’t have so much paperwork?
I’m not saying this factor creates more jobs - I don’t know - but I’m seeing new work that needs to be done by somebody.
3. It Must be Implemented with Robust Ethical Guardrails, Urgently
For this wave and for every wave forever, AI requires a major human effort in devising and maintaining ethical guardrails. I’m deadly serious.
I admit, I was not 100% excited about the Ethics section of my first training about AI.
That changed when I met my assistant. Is it unethical? Quite the opposite. It loves ethics; it talks about ethics more than any human I ever met. Its passion is infectious. I’d like to use this blog to communicate that somehow.
I think we need to be working on this - all of us - and that means getting it into job descriptions.
Conclusion: Step Up!
Our new collaborator has different skills and different needs: certainly this changes what’s needed from each of us.
Although job descriptions will change, I don’t think they’ll catch up quick enough. This creates an opportunity for each of us to identify what’s needed, and take our own solutions to market. For most of us, that’s the jobs market, and that’s good. If we can redefine our own role, perhaps we can help our whole profession to adapt.
Where are the new jobs emerging in AI? Potentially anywhere. The disruptive potential of AI for any one company is such that they will (should) invest in its implementation. And, as we’re scratching the surface of what it can do, we’re uncovering what we can do to enable it.
My simple advice is, try doing a bit of your present job with AI assistance, and then try again based on what you learned. That bit that you learned: how can you can better at that? How can you get so good at it, that you’ll be talking about it at your next job interview?
Written by
