Discover more from Experimentation Labs
Why AI isn’t taking over anything anytime soon
Not your job, not Google’s
Hi 👋 and welcome to Dan’s Growth Newsletter, where I try to help you grow your business. If you like this content, subscribe (it’s free)!
First of all, I didn’t, and still don’t, really want to write this newsletter about this topic. Mainly because I’m so sick of every article, podcast, and YouTube video talking about Chat GPT and DALLE and all the machine learning/artificial intelligence stuff that’s going on. But my wife said my hot take was worth reading, so here we go.
I like machine learning and AI advancements
First, let me say, as someone with a little bit of a computational linguistics background and a lot a bit of an SEO background, I think some of the improvements in these programs are super cool. I think they’re fun to play with, and the future is cool. So just know that as I start to rail on these technologies, I’m mostly talking bad about the misinformed futurist.
Search is based on trust
At this point, everyone has heard something about Chat GPT and Bing and Microsoft trying to implement chat into search and really doubling down on it, so I want to talk a little about the fundamentals of a search engine.
Your usage of a search engine is 100% based on how much you trust that the search engine will give you the best results.
Why isn’t your primary search engine Bing or Yahoo? They probably have most of the same websites in their index. Why is Google the search engine of choice for almost everyone?
Because Google, at its core is focused on giving its searchers the most accurate results it can.
Google search’s entire business model is centered on having enough users that they can show ads. How do you get users? Build the best search engine.
If Google started giving you results that didn’t feel accurate, it wouldn’t take you long to start exploring other options.
This takes me to the next point which is:
Microsoft has everything to gain and Google has everything to lose
Microsoft currently has an estimated 2.5% of the search market share. That isn’t nothing, but it’s not much either.
So if Bing decides to launch a pretty edgy (and popular) new feature, they probably aren’t very worried that they’re going to lose market share if it tanks. But if the popularity and virality of the feature brings users to the search engine, they’re winning.
Google can’t afford to make the same move, at least not with as much risk as Microsoft is assuming.
If tomorrow Google releases Bard (their version of Chat GPT) and it gave bad, inaccurate and dumb results, Google could legitimately stand to lose users. Google can’t afford that mistake.
Google understands how the publishing world works, and this Chat model breaks that
Right now Chat GPT and other machine learning tools are scraping the web and collecting information. It’s a real ask questions later policy. So the results it gets just come from the internet (we’ll talk about factual accuracies in just a minute).
Let’s use Forbes as an example. If Forbes writes an article they expect that article to make them money from affiliate links, paid ads and any other monetization methods they might have (membership, subscriptions, exclusive content, etc.) If Google and Microsoft are focused on a chatbot that just gives searches the answers, Forbes won’t make money anymore. If Forbes stops making money, they won’t write articles anymore. If they stop writing articles, the chatbots don’t have and source for their data. Are you seeing the problem here?
If Google introduces a chatbot that bypasses publishers ability to make money, then Google is effectively killing its business.
So in its current form, the internet as we know it is not set up to support the introduction of machine learning chatbots. As a gimmick its cool, but for these types of technologies to truly be successfully disruptive then there needs to be a dramatic change in the way that online publishers make money.
The results have been pretty inaccurate
So far the results of Chat GPT have been almost laughably inaccurate. In my experience you do just as much, if not more work to validate that what Chat GPT is saying is correct.
So if you spend just a minute getting an answer from Chat GPT but then 20 minutes validating the info, aren’t you almost always just better off figuring it out on your own?
There are ways that the results can become more accurate, but if my uncle can’t tell the difference I have a hard time believing a chatbot can figure it out.
Everything has inherit bias, Chat GPT is no different
Like I said, these chatbots are modeled on data they scrape from the internet, so accuracy is only one of the concerns. The other is the bias in the results. You can approach this from any direction, but if your chatbot has bias, the information you consume is biased, which is already the case for most social media.
How often could you predict your next day? AI can’t do it better
I don’t know how predictable your day is, but personally I vary my routine all the time. Siri used to give me suggestions on my phone that were so wildly inaccurate I turned the feature off. The idea that any AI or ML tool could accurately predict what I want at any point is fantastical to me. For an algorithm to predict with any accuracy what I want, it would have to study every second of my life for years, and even then I might still change my mind.
So the idea that AI/ML would be offering suggestions or doing things for me that are useful is laughable.
Here’s an example:
I have a garage that opens with an app on my phone. For some reason the app offers me the opportunity to set up routines. I haven’t explored this feature to its depth, but the idea that I would set up a routine on my garage door opener is insane to me.
What if I wake up late? What if I get stuck in traffic? What if I’m not going home first? There are so many variables that are introduced every day in my life that trying to systematize and program routines around what I do is honestly impossible.
Your job is probably safe for now
We don’t have flying cars. We don’t even have self driving cars. My phone can’t recommend the right apps during the day.
AI and ML are very cool technologies that are progressing really fast and have the potential to do some incredible things. But anyone who tells you that those technologies will change your life or take your job in the next few years, they probably own stock in the company.
Now I don’t want to be perceived as a total Luddite, because I get it, the internet came and changed all of our lives. Over the course of 40 years. But Bitcoin hasn’t really changed my life, and neither has the block chain. At least not yet.
I honestly believe that these technologies have an incredible future, but we are not there yet.