Navigating the Challenges and Opportunities of Artificial Intelligence in the Age of ChatGPT and Machine-Generated Content
ChatGPT has the enormous merit of having brought the topic of Artificial Intelligence to the center of public debate. A cold shower that suddenly woke us up from a torpor we were falling into and put us in front of the evidence that this family of technologies is running fast.
ChatGPT once again show us the importance of interfaces: it becomes immediate to talk about “the importance of AI” and to realize that you can’t wait to implement it in your company, when you have in your hands a little game that looks like WhatsApp and Telegram that answers any question.
I don’t know if OpenAI will lead the future of AI, and I tend to downplay the rhetoric for which Google is dying, but I think this “coming out of the closet” will open up an extremely interesting season of chasing, one of those movements in which the ability to be more vertical and specialized will give rise to many successful new companies at a time when the market has very high attention on the topic and huge needs for efficiency and renewal.
I’m also optimist on the side of content production, which is so much talked about. The first thing that comes to mind when the barriers to entry for such tools collapse with this speed is, “the Internet will be filled with crap contents.”
With some friends we are doing a thousand tests to connect the dots and see where automation can go, and we have come up with some very interesting things as well.
When our software automatically wrote and published a review of a mountaineering shoe in trend (with trend identified completely independently) in which cowhide inserts were referenced, we were initially disappointed with the quality of the information, knowing the brand’s focus on environmental issues. We went to sleep unplugging the software and saying, “is this really the desirable evolution for the Internet?” The next morning, searching more thoroughly, we discovered that the name of the leather used, which sounds like a synthetic leather name, is actually bovine-derived. Was “Bovine” the highest probability term after “Leather,” or was it actually the highest probability term in that context for that very reason?
If I go back to a 2020 Google paper, “Generative Models are Unsupervised Predictors of Page Quality: A Colossal Case Study” it comes naturally to “predict” what would have been Danny Sullivan’s assertion about machine-generated content penalizations: this huge quantity of content will likely help algorithms more accurately define and identify quality.
Yes, this flood of content finally makes everyone see that the emperor has no clothes. There may be machine-written content with more human reasoning behind it than a lot of content written in an assembly line composed of people… and that perhaps this will finally be the time when we stop evaluating quality based on “word count” or with the immortal myth of “keyword density”.
This period of acceleration will be a period of great opportunities and challenges. Opportunities and challenges of an entrepreneurial, professional and ethical nature — with a crazy mixture of the three perspectives. We have all seen how “the Oracle” ChatGPT turns relative concepts (found on the web) into absolute concepts with disarming simplicity. As SEOs, we know how manipulable such systems are (here’s a nice article that came out on AnalyticsVidhya a few days ago, but we’ve all experienced ChatGPT’s little syntax errors or code snippets copied from StackOverflow these days, just as we’ll remember the case of Google Bombing on George W. Bush).
We are facing the accelerating adoption of tools that will not replace us, but are extremely powerful and will introduce opportunities, challenges and vulnerabilities at the same time.
There will be a lot of work to be done, let’s keep our brains working 🙂 🙂
