Machine Learning will be Incredibly Disruptive
A bigger change to political campaigns than TV
In the high tech arena we have had three fundamentally disruptive technologies: the personal computer, the Internet, & the smart phone1. We now face the fourth with Machine Learning (ML) and it will be even more disruptive than the previous three.
No, Crypto is not a major disruptive technology. Crypto is a Ponzi scheme wrapped up in some cool technology. And blockchain is like the first 20 years of the laser, a solution looking for a problem.
The top news right now on ML is ChatGPT & DALL-E and what you can do with them. And both show so truly game changing possibilities. Already the High School essay homework problem is obsolete.
However, I just returned from Microsoft’s Azure + AI Conference and what those are showing is just the start of what’s possible now. Not what we might be able to do soon, but what can be done now.
For example, Microsoft has a service that listens to a Doctor’s examination of a patient and from that creates the case notes (charting). Not a transcript of the discussion - the case notes. This is easily the worst part of being a Doctor or PA and ML applied here means less burnout and therefore more Doctors2.
Something key to understanding ML is it takes a couple of weeks to develop a model for a problem. And it then takes a couple of years to train the model. So we have all these upcoming applications that are presently being trained. And once trained, their impact will be gigantic.
Political Campaigns in the Future
When the smart phone came out, no one predicted Uber, much less Twitter. How could you? But it turned out that being instantly at hand, broadcasts of short messages resonated with people. And so Twitter became a central part of our society and a key element of political campaigns.
So, let’s get into the effect of ML of political campaigns. What will be the giant changes, akin to Twitter? I have no bloody idea. No one does. Most of the major impacts will be something new, just as Twitter was something new.
But there are some clear avenues that will be of impact. ML provides three big things. First is prediction. A properly trained model will provide the most effective place to focus your resources from the candidate’s time to advertising to volunteer work. You’ll know the most effective way to reach each voter.
Second is as an intelligent assistant. You need a script for an ad where you want to get across a specific argument. Services exist now to write the initial draft for you. You’ll still tweak it to get it perfect, but creating any content you need has just gotten 5 - 10 times faster. And it’s now two of you writing it and the ML will contribute some brilliant phrasing you would not think of.
Third is to identify trends. You want to be campaigning in August on the issues that will be front of mind in October. Even better if you can identify issues that will be major in October that no one is talking about in August - you can then own that issue.
So focused fine-tuned targeted efforts where in many cases each voter is receiving a unique set of content with a unique balance of delivery mechanisms. What’s that worth compared to an opponent who does not make use of this? Probably a couple of points in the election.
With that said, I think there will be several significant new services or apps that come out that do not fall into the above categories that will be gigantic. So it’s critical that campaigns keep their eyes open to anything new, especially if it’s different, and be first to use it if it’s impactful.
The Downsides
And there’s going to be some major downsides. First off will be deep fake videos, audio, etc. of candidates. Prepare to see porn movies where one of the actors face has been changed to be a candidate. This is going to be especially awful for female and LGBT candidates.
But it won’t be just porn. It’ll be fake videos of say Biden receiving a handful of cash from a Russian operative. It’ll be fake videos of say Harris talking to a group about wanting to have the government force Critical Race Theory studies in Kindergarten. And the clever ones will be ones that work from a factual event and take it just a bit further.
The fake newspaper situation will explode. Instead of an online paper being 10 or so articles, with ML you can create an equivalent to the Denver Post in a couple of hours, with accurate articles on the regular news, sports, entertainment, etc. And then centered in all that credible content, the front page fake news story3.
You’ll also have opponents that will use ML not just to build up their candidate, but to tear down yours. If they can hammer you, and you’re merely responding, then they’re inside your OODA loop and that provides them a giant advantage. This could well be the most critical & impactful issue in campaigns going forward.
What’s Key
So what to do? The biggie is evaluate what’s available and try those services that look promising. Experiment on them. Don’t make big bets at first, make lots of small bets. And measure the results.
If you find 30% of what you try turns out to be worthwhile, you’re doing great. If over 50% turns out to be worthwhile, you’re being too careful and missing some game changers.
The other part is to be adaptive and respond quickly. And pay close attention to what others are doing, not just your opponent but other candidates. It’s those unknown unknowns that can really bite you in the ass.
Finally remember that in times of great disruption, there is tremendous advantage to those that embrace the disruption and take maximum advantage of it.
In Biology we have mRNA and CRISPR. In energy we’re approaching abundant inexpensive renewable electricity. The list goes on…
This may end up significantly helping legitimate news media. If every news item, every newspaper, every video is questionable and could well be fake, then trusted sources become a lot more important.