I occasionally visit Costco, and always with trepidation.
The prices are amazing, but the volumes are always more than you need. I wonder whether these “deals” end up costing more over time as “big box” food purchases, for example, appear to increase consumption.
When is a deal a “good deal”?
It’s a nuanced debate that I have with my family around potential Costco visits (especially with the impending festive season).
Which brings me neatly (?) to Agentic AI and Autonomous Buying . . .
Agentic AI describes a technology segment that can perform complex tasks autonomously with minimal human supervision and can;
- Comprehend and follow complex instructions
- Set goals independently
- Plan workflows and adapt based on changes in circumstances
- Make decisions
- Learn and improve from experience
I read an interesting article on Agentic AI and “autonomous buying” (which reminded me of my Costco experience 😉 ) by Phil Wainewright of Diginomica.
Fortunately, my hyperbolic image of AI agents frenetically buying up all the “deals” for the enterprise and thus putting it out of business, was met with a more common sense vision of human augmentation.
“Autonomous agents managing the objects, interacting with buyers and sellers, and making appropriate recommendations as needed. . . . . . .”
I felt reassured with this vision of AI in spend management given my own experiences in trying to focus “Human Intelligence” (HI) and judgement on “when is a deal a “good deal”?”.
With a broader perspective on Agentic AI, Bernard Marr wrote an excellent article in Forbes.
“At its core, agentic AI refers to artificial intelligence systems that possess a degree of autonomy and can act on their own to achieve specific goals. Unlike traditional AI models that simply respond to prompts or execute predefined tasks, agentic AI can make decisions, plan actions, and even learn from its experiences – all in pursuit of objectives set by its human creators.”
Of course, we have all seen the challenges of AI algorithms “in pursuit of objectives set by human creators” writ large in the social media amplification of unusual and surprising “facts”.
Agentic AI also incorporates a “chaining” capability that can take a sequence of actions in response to a single request, breaking down complex tasks into smaller, manageable steps.
Bernard rightly comments on the decision making challenges ahead with Agentic AI, especially with regard to human values, data privacy and transparency.
“The complex nature of AI models can make their decision-making processes difficult to understand or interpret. This “black box” problem poses challenges for accountability and trust, especially in high-stakes applications.”
The lack of transparency behind the “objective function” set by human creators is a big element in this “black box” problem, as we continue to see in the results of social media algorithms referenced above.
“There’s also the question of accountability—who’s responsible when an agentic AI makes a mistake?”
Bernard concludes with a strong reference to the “augmentation vs replacement” debate.
“The key to harnessing agentic AI’s full potential lies in striking the right balance between autonomy and human oversight. By developing these systems thoughtfully and with a keen eye on ethical implications, we can create AI agents that augment human capabilities rather than replace them.”
Thanks for reading . . .
Closing thought – Will Agentic AI be able spell better than GenAI?
