AI is already changing how users and content find each other. It means significant changes for both search and content creators.
Imagine you are a newspaper editor in the early 2000s. For decades your paper has done well. You have good writers, good print and excellent credibility with your audience. However subscriptions don't show it. Slowly more and more people are shifting towards digital news. However you are hesitating to switch because it is still difficult to browse mobile. Responsive web is in it's infancy, access to internet is mostly limited to advanced and expensive devices, and people are not spending much time in internet outside of their homes. Jump ahead a few years and in 2007 iPhone comes in and revolutionizes the way we can consume digital content. Suddenly the rails are off and there are no limitations for consuming content online.
I don't think I am being bold when I say we are in a similar moment in history. AI is about to revolutionize how the content is created, how we search and in turn affect everything that content creators and publishers have learnt about the web.
There are so many unknown possibilities of how this can play out, so understandably content creators are worried about the future of their craft.
It is no secret that AI has found its place in every step of content creation. It can help you with ideas, research, outlines, editing and rephrasing, even write a full blog post. Businesses are starting to use AI instead of copywriters whenever they can.
And it won't stop. There will be more and more AI content filling every corner of the world. And the society will adapt fast, much faster than it did with the initial digital boom.
AI content is good enough to pass as readable for general non-reading population however the quality is not excellent by any means. Regardless, it does the job well enough not to ignore it.
In addition to the increased competition, the landscape of search engines is also changing. AI assistants can help you find an answer to any problem, without you ever opening a single website. As a result users will spend less time searching, and the amount of overall traffic to get to the same answer will decrease.
Less traffic means less ad impressions and clicks. For content creators a decrease in traffic most likely signals a decrease in revenue.
Informational content will be affected the most because AI can replicate and compete with that type of writing. E-commerce is likely next on the list, however a lot more complex. Let's say you would ask AI to help you find the best flip flops for your vacation. Businesses selling flip-flops will fight for the chance to be referred by AI. As long as AI is using the web to search for realtime information the fighting groung is still the traditional SEO. But as AI keeps advancing new areas of competition are likely to emerge.
AI is competing with content writers, content writers are using AI to compete with each other, and AI is exploiting the web as we know it.
Not that web wasn't already filled with spam and useless articles. But should you be worried that everything in the web will be AI generated eventually?
Google does not seem to care, as long as the content is useful to the user. For Google, useful means original and high-quality content that demonstrate the following qualities E-E-A-T:
With AI tools such as ChatGPT getting better and better, it will become even easier to generate content that seemingly can satisfy these requirements.
But let's be clear, given the current models such as GPT-4 you still need careful planning, review and modifications to make the article truly readable. There is nothing entertaining about AI spitting out and rephrasing the same information that has existed for a long time. There is still a human element needed.
At least for now, until a more advanced AI emerges... 😟 (AGI)
Besides the fact that Google is focusing on delivering valuable user experience, another important reason why Google does not punish AI content is the fact that it can't reliably detect if text is human or AI written.
AI detectors are always in a chase, trying to catch up to newest algorithms and generation models. Most of the existing AI detectors can identify simple, unmodified AI content. However they have real trouble with AI content that has been rephrased using AI, not to mention mixed with some good old fashioned human writing. In most cases to bypass the detectors all you need to do is generate AI content, and then have another AI prompt rephrase it. No human touch needed.
There are some claims that Google can, and simply decides not to detect this content, however this is an outdated view that no longer applies to modern LLMs. In reality, there is no basis for such a statement.
So, even though Google is saying that they don't punish AI content because they see it as useful, the fact is that they couldn't even do it if they wanted to.
Let me give you an another example. In writing this article, my purpose is to give you my unique perspective, something to make you think, or at least see what I am thinking.
As always before writing an article I research the competition. At the moment of writing the draft name of this piece is - how will chatgpt affect content creators.
The first result is clearly AI generated (even the not so advanced AI detectors can see that). As a human with general interest in the topic it is difficult to get through that article. Good job EEAT.
Perhaps if you are reading about ChatGPT and AI for the first time, you might be able to learn something but in reality it is repeating common knowledge. I asked ChatGPT to summarize the article and to give it's opinion:
While informative, these topics do not appear to present unique or unknown information that couldn't be gathered from a fundamental understanding or a simple inquiry about ChatGPT's functionalities.
Well, yeah, it is a pretty boring and pointless article. But somehow Google thinks this content is helpful and positive search experience.
Before the launch of ChatGPT I would have agreed that this article is useful. But now, when AI is so accessible, anything that ChatGPT (or AI in general) can answer should not even be competing for a spot in search.
The content in search should be unique and truly useful, or thought provoking. There should be no reward for reciting the same old knowledge.
There are services that help people generate unthinkable amount of articles with almost no effort at all.
These services use a strategy to generate thousands of high-quality articles to target low competition, low volume keywords that would otherwise be cost inefficient if written by a human. By using AI to generate the articles they are able to generate thousands of articles instantly, with very little cost. Even though the keyword volume might be low it is the scaling that allows them to grow the traffic significantly.
It is a strategy that works well because of the low competition. For now, there is nothing stopping you from using this strategy. It works.
There are a few things that bother me about it. For some reason, Byword (one of the services that provide AI generated content) in their blog insists that Google is simply choosing not to detect AI content.
Google isn’t playing technology catch-up with OpenAI/language models. It’s been fairly easy to detect AI content for some time. / Byword.ai
Their only source for this is a model from 2019 RoBERTa Base OpenAI Detector. The model itself admits that it is not reliable and should not be used for any decisions. Not to mention it was built for GPT-2 models. I hope this might encourage someone to update this page with some relevant content.
I do agree with the next sentiment:
Google has no business incentive to care how you made your content, only that it contributes to a positive search experience. / Byword.ai
I agree, I completely agree with that. What I don't agree with, is that these long-tail keyword knowledge bases are creating or enhancing the positive search experience.
AI has the potential to do it by revolutionizing how people search. Not by hacking legacy SEO techniques.
Furthermore, let's ignore the scenario where a lot of people start competing for these low volume keywords, what a beautiful sight will that be (probably already is).
What happens when search engines truly embrace AI? What unique content do these lon-tail articles have, to warrant even a consideration to be read? As we saw from one of the previous examples - nothing. Most of those articles are regurgitated information targeting less popular keywords to scale traffic. Still, I don't doubt that it works as as short term strategy.
If we look at it from a bigger picture perspective these articles should not even exist, they are simply a waste of data center resources, waste of AI resources and most importantly waste of time if anybody reads them.
My hope is that AI will help users formulate their thoughts, resulting in search box/ keywords playing a much smaller role in search.
So, should AI content be punished? In my opinion, no. Especially since it can't be reliably detected. It would only be a half-measure.
Instead search needs to double down on it's values.
I think Google is on the right track with defining their values. Regardless of the source, content should be unique and valuable, and helping the user find what he was looking for.
Does that mean that Google has it figured out? Not yet.
Just take a look at almost any Google result, for example, let me search for ai content generators.
Be honest... how great do you feel seeing these unique and high-quality articles right now?
It is painful to scroll through these. The first few pages of results consist of websites competing for keywords. Most of the time they do it by copying what Google is rewarding (their competition) and adding some slight improvement in hopes that Google might reward them the top spot.
I love that the first on the list (best AI generators) is always their own AI generator. A whole industry built around duplicating, copying, regurgitating and trying to get your share of the pie.
Competition raises the bar? In a way - yes. Although I would not say that the first place article is much better than the others. Plus I would argue that some high quality topics are hidden somewhere deeper in the results.
In a way, content on Google already is a bit like mass produced cluster of similar articles. Does it matter that previously these were written by humans with little to no expertise on the topic? It could just as well be written by an AI.
But the biggest problem with current Google search is that it is based on the users ability to use the search box. The search box has improved over the years, providing popular searches, finishing a sentence in some cases, filtering,etc... However at it's core search box is so static, so plain. An everyday person does not have the skills to easily describe what they want to find. Sometimes you don't even know what you want to find, you are hoping to learn something from what you are shown. And then you spend hours trying to figure out what you are actually looking for. Or what keywords you need to put in to get the information you need. Even if you do know exactly what you are searching for you are at the mercy of the publishers and SEOs understanding of what each keyword means.
This is where AI can step in, and change search experience entirely.
AI can understand the context of what you are doing, it can help you formulate what you are looking for, it can help you show different solutions for your problem. And the best part is that you don't need to waste hours walking through copies of (almost) the same article, trying to find that one detail that you were missing. AI has done that for you.
And if you are interested to know more, AI can reference the source.
So how does it affect content creators?
Obviously, in order for AI to have somewhere to get its information, it needs content. It already has a lot but it will always need updated and relevant content.
Training AI using AI articles would be pointless for the purpose of providing unique value to the user. It needs up to date information. So content creators will be needed, period.
One area that might completely change is how content creators are rewarded for their efforts. Having AI to summarize and help find the exact thing you are looking for will mean that search traffic should reduce significantly (*not counting AI traffic because AI does not look at ads - or could it? 🤯).
Assuming that traffic is lower, there should be less ads shown. Less ads shown means less real-estate to put your ads on. I doubt Google is going to be the losers in this scenario, they can simply increase the cost of ads. Businesses will still need to advertise, so they will pay and try to get the now much fewer spots to advertise on.
However, as AI search gets better, it will easily weed out the duplicate information, the articles that add little to no value. For example, if now, in the current search, the article in position 8 is getting some traffic from users trying to find a specific detail, it might get no traffic if there is nothing unique in the article to make the AI reference it.
As a content creator you will likely get less traffic from search but assuming your content has value you should still be rewarded similarly given the higher cost of ads. This does assume that AI responses would include references to your blog.
Which article would AI reference in a case where that they both have the same information? I don't know. Perhaps it would A/B test some articles to see which one solves the curiosity of the user better.
One thing that I like about this is idea is that there would be less incentive to copy or cook up an article same as your competitor. It would still happen but hopefully much less. Definitively less from those whose only business is traffic.
Ok, so now we have lower overall traffic from search. Unique writers who benefit from adsense might still keep some of their income. Lower quality writers might go from something to nothing.
Is this the only way? Is there nothing the content creators can do?
Publishers fighting back and blocking AI crawlers
Some publishers have opted to block AI crawlers and not allow AI to use their content in their models. The reasoning for this choice is simple. The fear of AI using your knowledge to answer the questions directly, therefore making sure the user has no reason to go to your website.
I won't lie - it is a completely realistic scenario. There are no guarantees that AI would ever refer the user to your site, and no reason to think that you would get any consistent traffic from AI referrals. However, I doubt it would happen that way. Bing is already including references:
It is a big risk. By blocking the crawlers you are potentially missing out on a huge opportunity.
Instead of blocking, a strategy change might help you more. Content writers that rely on traffic can't continue to assume that search engine rules won't change. What happens when Google transitions to their own AI? Would you block Google as well?
A simple strategy would be to diversify your income channels. Implement a membership based website where you give your audience what they want. Give them a better experience than AI can. Use Google and other AI search platforms to grow your audience. You don't have to show AI everything. Just enough to get the referral. AI can work for you as a promoter but your uniqueness is what will continue driving users to your site.
If you don't have anything unique, or anything that people might return for, then should you really be writing for a living?
I think memberships and newsletters might become much more popular. Once people realize AI is not a person, they might start longing for that human contact in the form or reading.
As an example, would you truly be interested in a AI social influencer? Maybe for a joke, and just because it is new and exciting. But really, it is just like following a toy character with a made up personality, while entertaining you probably won't care much if it stops posting.
In addition, blogs are supposed to be entertaining, regardless of what topic you write about. People read and stay because of the writer, not the facts or content. At least that is what I feel writing should be about. AI can not replace your personality, your unpredictable ability to keep the user interested.
My advise - don't block the AI crawlers, instead evaluate your strategy. But if you have no strategy other than search traffic then start looking for another job. When was the last time you bought a physical newspaper?
At the moment it is clear that AI still relies on the existing web structure and search engines. Instead of just enhancing the search experience, can AI completely obsolete the search as we know it?
Given how the web is built, in order to compete with search engines ChatGPT would have to build a search engine itself. Crawling and indexing the whole web is a huge task, even for LLMs. Especially given the fact that every day millions of AI generated articles are published. Google (and Bing) already has this part covered, it would be an amazing achievement to overtake them.
But theoretically, web might change in the future. Instead of making websites that are found via search engine, we might be creating APIs that talk with different versions of GPTs. Content creators would be directly talking to AI and getting some reward from this process.
It is quite likely that the current web is some caveman version of what it will be in the future. We still have a long road ahead before we can see what is the next step in this evolution.
Regardless of the long game, it is clear that changes will keep happening and the days of legacy SEO might be coming to and end.
My inspiration for the article: