9 hours ago
Using AI Correctly May Take Some Thinking
Digital generated image of young woman standing on red background and putting hand into multicolored ... More glowing portal door. Metaverse concept.
To many people with an inside view of the industry. It's the details of our work with LLMs that matter. If you're sufficiently involved that you already understand the major applications and use cases, then you're likely to be thinking about exactly how this will work in any given business.
A CEO is in the hot seat in this way. He or she has a certain responsibility to know what the trends are, what the big technologies are, and how to deploy them in a general sense. He or she also has the responsibility to apply them to the exact business processes that leadership is in charge of.
Some of this involves understanding how people work with processes and products.
I hear a lot of people use those 'three P's': people, processes and products, to talk about the overall environment in which we're figuring out how to use AI.
Here's an interesting analogy: think about it this way – buildings have doors. People need to go in and out.
So sometimes the door is made for humans to push on it to get access. Other times the door is made for people to pull on, in order to get in or out of the building. Then there are revolving doors.
The commonality is that people need to know how to use those doors effectively. If you're pushing on a pull door, you're not going to get anywhere. And vice versa.
I heard a lot about this kind of challenge in a panel at Imagination in Action this spring.
Panelists talked about various challenges, and Kevin Chung went first with ideas on employee alignment.
'Employees are sabotaging AI strategy,' he said, estimating that a full 40% of employees at a given workplace are 'not aligned' with what leadership wants to do. He also stressed the importance of AI literacy, which faces challenges, in light of the fact that these technologies are only a few years old.
Panelist Hira Dangol talked about business alignment and ROI.
'I think that governance is one of the key enablers, and also a limiting factor when it comes to how you are going to achieve success at the transformation of your AI journey,' he said. 'And then if you look at the value chain structure from the bottom up for the enterprise, it's pretty much like a buy versus build model.'
Noting the utility of events like hackathons, Olga Beregovaya described a set of analytical processes for vetting AI.
'It works great at the lab, but how do you actually take it to enterprise level?' she said. 'How do you make sure it's scalable? How do you mitigate technical constraints like, say, latency, when you need to deliver content instantly? So it's basically how do you manage supply chain and global content delivery chain, end to end, making sure that AI has a role where it really does have a role.'
She also talked about human translators being involved with AI's treatment of language. So much of the technology, she pointed out, is English-driven, and other languages can pose a challenge. In some ways, it's amazing how AI adapts. In other ways, there are still gaps and hurdles to get over.
'Now, when AI … works great despite those 20% of hallucinations and other flaws, what do translators actually do?' she said. 'And how do they reinvent themselves? And that touches upon the ethics, right? You want to make sure that people have jobs, and you want to make sure that workforce emotional safety is at play there.'
At the end, each contributor shared their optimism and pessimism over AI.
Chung balanced a lot of interesting applications like self-driving vehicles with the idea that great power comes with great responsibility.
Dangol talked about security, privacy and risk management standards.
'It works,' Beregovaya said, again going back to cross-language capacity. However, she added, AI is not a 'silver bullet,' and people have fears around adoption that have to be dealt with.
All of this provides us with some good feedback for figuring out how to integrate AI to business in the right way. I've always said that technology can help or hinder a team based on whether it's the right fit or not. Let's make sure we have it right before letting these technologies transform enterprise.
