The Digital Marketing Podcast Ep. 327
~April 2023
Premise:
Google announced in 2022 that ‘you are not allowed AI content on your website’
AI-Generated content wasn’t clearly defined.
Now in ~April 2023 Google re-announced and said:
With the exception of AI tools which are designed to attempt to manipulate Google Search rankings, AI-Generated content is allowed
Google has published it’s own AI-Generated content tool ‘Google Bard’, so without this announcement Google made themselves prone to criticism of hipocrisy
Daniel Rowles believes that even with the power of AI, there will be a high bounce-rate if content is generated solely by AI
Data sourcing for AI models is important in terms of AI-generated copy, because the content has a high sensitivity for misinformation, for instance:
Referencing the brand name of a competitior
AI tools to note
TFIDF - Term Frequency inverse Document frequency → Read a document and note the frequency of a specific keyword/keywords in order to ascertain a specific relevancy score
Example services:
- SEMRush -
- SurferSEO - Ranking your written content based upon contextual analysis of existing, well-ranked content using AI analysis tools on the existing content
- Suggests perfect sub-headings within the articles
- In the realm of being a ‘keyword research tool’ to help you rank your content intrinsically whilst writing
- Here AI is not generating content, meaning high potential for content authenticity
Data sourcing:
Washington Post article - How many websites were accessed with private or otherwise copyrighted/protected material which AI models were trained upon. This does cause copyright and data protection issues
There is also vulnerability for AI manipulation due to blacklist being created for websites which contain disinformation (known, intentionally misleading or even propagandistic information)
There is not transparency from the models as to where the data was sourced from (because all the training data is aggregated)