It’s the First Budget Season of the AI Revolution
It’s the First Budget Season of the AI Revolution
How should newsrooms be thinking about that?
Welcome to Second Rough Draft, a newsletter about journalism in our time, how it (often its business) is evolving, and the challenges it faces.
I haven’t written much until now about the generative AI revolution for a number of reasons. First, it’s one of those subjects on which there’s been too much heat and not enough light shed, too much utopianism and dystopianism at one and the same time. Second, it’s neither a subject of my particular expertise nor one especially focused on journalism.
But I think it’s becoming increasingly clear, all hype aside, that this is the biggest turning point in technology since at least the Nineties, with enormous implications throughout our economy. So, while journalism may not be uniquely affected, it will surely be changed.
And that, given that it’s now July, the beginning of budget season for many newsrooms, brings me to the near-term budget implications, something about which I think I may know something.
Once upon a time
My thoughts, when confronted with yet another new AI capability, keep returning to 1992, when I assumed responsibility for budgeting in the Wall Street Journal news department. Under my capable predecessor, crafting the news budget had taken one senior manager most of every summer, with a raft of sharpened pencils a key tool. With the advent of electronic spreadsheets, and especially with the breakthrough of Lotus 1-2-3 Release 5 in 1994, which added the ability to automatically link and update multiple spreadsheets, the process, within just a couple of years, was reduced to four person-days of effort, two by a senior manager, two by a skilled assistant. No pencils were involved. Real-time tracking of the salary budget, a huge aid to efficient hiring, quickly followed, and the freed time made possible a host of new initiatives. In my world, that was revolutionary.
Here are a few observations about how current budgeteers might think about these early days of the generative AI revolution:
For those under pressure to limit expenses (which, at this point, means almost the entire industry), there are already some huge temptations. As you’ve seen in this newsletter in recent weeks, it’s possible to generate credible original illustrations quickly and at essentially no cost. How much longer can it make sense to pay human freelance illustrators for results that, in most cases, may be better but not materially so?
The same is true of basic copying editing (conforming to style, correcting grammar, doing a final fact-check—although not yet an initial such check).
For many reporters, AI can already substitute significantly, although certainly not entirely, for the assistance provided, in better-resourced newsrooms, by a dedicated research staff.
Many early-stage and time-consuming tasks in data journalism can be accomplished much, much more quickly. Like most of these efficiencies, this poses a question: use the savings to limit spending or redirect them to enhance capacity?
As in much of the budgeting process, all of this produces a series of hard choices. There are human costs to realizing some of these cost reductions. It’s easier to limit freelance expenses than cut staff, but still painful. Do you want to hold one or both of these lines for now? Can you afford to do so?
Opportunities as well as cost savings
Even if you aren’t under irresistible pressure to cut costs, the new AI may still create hard choices of another sort. That’s because it also offers amazing new capabilities. If you can afford flat budgeting, do you want to make savings where possible to free up resources to spend on some of these?
Two of the capabilities most immediately available involve creating new versions of already-crafted stories at previously unheard-of low cost. Such versions can be merely shorter, or designed for lower reading levels, perhaps opening a path to new audiences and younger readers. Alternatively, comprehensive versions can be created in additional languages or modes more accessible to those with disabilities. These things were, until this year, only within the reach of the richest publishers. Soon, they will achievable by all, and not offering them will constitute a choice of its own.
Like my quaint story about Lotus 1-2-3, you can be confident that this column will seem almost archaic in just a few years. You may have noticed, for instance, that I actually haven’t mentioned the possibility of AI writing stories by itself, because it’s mostly not quite ready for that, at least outside of formulaic pieces. Nor have I talked about what AI will mean on the revenue side, where I don’t think the picture is yet quite as clear.
Even more significantly, I haven’t said a word about new journalistic forms and forays. The real AI revolution will come to journalism when it makes possible not just old forms at lower cost but new forms altogether. Who will create the first AI equivalent of “Snow Fall.” and when? Exciting times, with hard new questions of which we haven’t yet dreamed, lie ahead.
There has been a lot of nonsense written and idle speculation about the use of AI in journalism. This piece is not one of them. It is well-reasoned, carefully measured and the author knows his material and when he is not certain about impact he says so. Of course the expected short-sighted comments are already expressed- instant, off-the-cuff ones that need to be taken with a grain of salt.
All these decisions will be and are decided by the interests of Capital, so these cuts to labor are guaranteed to come, the acceleration of wealth concentration and its problems for newsrooms will increase. AI is indeed the new McKinsey. One hedge fund alone controls many of our papers having just snapped up one in San Diego and they have investments in AI to further solidify control of industry and narrative.
That said, these communication technologies and the underlying foundation of free speech are the best hope to counter looming dystopia, but that requires inspired education, as of the kind MLK provided when he came to a town and got folks who used to demure to racism to instead face dogs, water hoses and death. This points to the notion that journalism must be more subjective if it is to protect the grounds of dignity - because that is the only place where objectivity matters.
AI will force the issue of Solidarity because the lack of it has allowed shamelessness to flourish. We are veering from the shameless into the reckless as Capital has run increasingly unchecked amidst a number of crises that may converge in our lifetime. If there is one thing we can be pretty sure of, humans -- especially ones reared to celebrate individualism and who fear education to such an extent they celebrate disinformation -- will not act in their collective interest until the menace is already upon them. A more subjective journalism can be a strident deterrent towards that kind of outcome.