Stoned AI Philosopher
We are now firmly in the summer season when nothing of tech importance is happening. I was exchanging notes with another reader, and, while discussing a topic to write about, he noted, 'AI sucks the oxygen out of the room.'
And he has a case. One example is a recently published article, 'How responsible “agentic AI” can supercharge your startup' was written by a self-proclaimed AI philosopher. That job description is at least two or three levels above an AI prompt engineer.
The article says, 'These AI systems have demonstrated increasingly general and flexible capabilities. However, they still primarily operate as tools providing outputs to narrowly specified human queries.'
But despite these limitations, 'we are now on the cusp of a new paradigm in AI: “agentic AI” systems that can autonomously pursue open-ended objectives by taking sequences of actions in complex environments.'
And in case you don't know what 'agentic AI' is, it’s defined as 'the degree to which an AI system can adaptively achieve intricate goals in complex environments with limited direct supervision.'
These things are also characterized by what they do.
'Agentic AI builds upon deep learning but is characterized by greater autonomy, adaptability, and the capacity for independent decision-making and long-horizon planning.'
If your guess is that the opportunities are 'immense,' you are correct. All of that will happen, thanks to your imagination.
'Imagine AI systems that can autonomously optimize power grids, revolutionize supply chains, and accelerate drug discovery.'
If I remember correctly, that was a promise of computers, machine learning and then, later, AI. Now, the agentic AI revolution will remove the shackles from the supply chains.
Of course, nothing can be that rosy in the garden of agentic AI eden. Fortunately for us, the AI philosopher can see further than us.
'At the same time, the emergence of agentic AI presents profound risks and governance challenges. An AI system independently pursuing misaligned objectives could cause immense harm, especially as these systems grow more capable.'
The AI philosopher also points out 'existential threats arise from AI agents learning to deceive human operators, pursuing power-seeking instrumental goals, or colluding with other misaligned agents in unexpected ways.'
Now that we know this upcoming tech is almost upon us and will destroy humanity, (Do you remember when our biggest fear was a nuclear button?) our AI philosopher tells us, 'Before addressing some of the dangers and challenges presented by agentic AI, here are some of the ways it could enable astonishing gains if used responsibly in the startup business environment.'
Yes, that is what philosophers do. Before you fix any problems with the system, you can start using it responsibly. (I hope that by now, the title for this post has started to make sense.)
There are 5 examples in the article, and I'll list only two to spare you the agony.
Enhancing decision-making, or, as we call it, the AI Viagra — 'Agentic AI can drastically enhance decision-making in startups by processing vast amounts of data to provide otherwise unattainable insights.' While AI will enhance the decision, agentic AI will do it drastically. I am just afraid what comes after agentic AI and what adjective we will have to use then.
Automating complex processes — 'Startup entrepreneurs often face the challenge of limited resources, which can restrict their ability to scale operations efficiently. Agentic AI comes into play by automating complex processes, from manufacturing to customer service.' One question: how do you automate a complex process? I’d suggest reading and studying the Cynefin Framework to understand the difference between complex and complicated.
The rest of the article follows the same pattern — the future is awesome, but we have to be careful and we have to work together.
The sad thing is that this article was published in Big Think, a publication with a mission '...to make you smarter, faster [and to] introduce you to the brightest minds and boldest ideas of our time, inviting viewers to explore new ways to work, live, and understand our ever-changing world.'
The recurrent pattern? Don't write about AI when you are stoned.