Aug 27, 2024
That modestly articulated value proposition/tagline for Google’s Gemini for Workspace is easy to understand and hard to deny in terms of appeal. What employee doesn’t want to automate away the drudgery and be free to focus on the most interesting/rewarding aspects of the job? And what employer doesn’t want to boost the productivity of their people?
As dreams of an AI-powered future of work go, the one focused on increasing productivity and augmenting the capability of employees is definitely among the more broadly appealing and human-centered – especially as compared with persistent C-suite fantasies of automation-enabled staffing cuts made in the name of efficiency and profit margins. It’s also a dream big AI players have leaned into this year as they hustled to package their respective AI models as productivity-enhancing smart assistants, no doubt hoping to counter some of the more critical takes on the productivity promise of Gen AI tools that were beginning to bubble up in the wake of the sustained hype wave of 2023.
To a certain extent, this has been a success (although we at Radical have been a little skeptical of some reports of worker adoption of Gen AI tools), and there are some interesting signals pointing toward a near future of AI super-empowerment extending well beyond the automation of busy work for entrepreneurially minded workers – perhaps especially those at early-stage startups and scrappy small businesses. Research from Gusto suggests that more than 20% of new businesses in 2023 were using Gen AI tools to augment capabilities, and that figure has surely increased for 2024.
So far, so good. But let us offer a word of caution here: With that promise of enhanced capabilities comes an increasingly high risk that organizations and individuals will promise value that they can’t deliver or support beyond the surface level.
Consider an example from our own experience: A little while ago, we were approached by an organization to deliver a three-day leadership development program for one of their clients. A detailed agenda had already been created, and the end client had approved it. The catch? The program agenda seemed to have been developed with very little awareness of the existing capabilities and content library of the seller organization or those of any potential third party (e.g., us at be Radical) that might be contracted to help deliver the program. This was doubly problematic because the timeline was tight and the agenda was highly specific. That combination of high specificity and low connection to the reality of things on the ground (along with some telltale formatting and language quirks) had us strongly suspecting that the agenda was a Gen AI product created just to get the client to yes. Anyone who took on the responsibility of delivering that program was either going to have to build it from the ground up in a hurry (possibly by leaning heavily on an LLM themselves 🤔) or would be stuck navigating an awkward conversation when the final deliverable didn’t look much like the program the client had agreed to purchase.
Wildly bullshitting to get a client to yes is, of course, not a new thing. But make no mistake: The supercharged ability to wildly bullshit in an era of instant pitches, proposals, presentation decks, and business plans, well… that has the potential to create some real chaos. And the LLM will create what you ask it to create (perhaps engaging in some bullshitting of its own) without any regard for whether you can actually deliver the thing it’s helping you pitch.
Now consider that the Gusto research I cited above also noted that “marketing is by far the most popular use case, as 76% of new companies that are using generative AI are applying it to marketing tasks.”
The moral here, friends: Use your AI productivity superpowers wisely. Keep a human in the loop. Be sure that your teams do the same. And of course, let the buyer beware. (And the recruiter too.)
via @Jeffrey
