1 of 44

2 of 44

Strategie SEO per contenuti a prova d'intelligenza artificiale

Simone Righini - 2022

3 of 44

Dalla pratica alla teoria

4 of 44

NLP

Natural language

processing

LLM�Large language models

5 of 44

Predictive AI solutions are MLs and there's no reason to think that they cannot be predictive.

The question is HOW predictive and at what error rate. To meet the minimum standard for "working" they only need to be sufficiently better than tossing a coin to justify their cost.

The problem isn't inherently in the tools, but in the deception around how the tools are marketed. These predictive ML models are not all-seeing god-look oracles.

But that's exactly how they're marketed. And because NOT A SINGLE one of these companies will permit their tool to be subjected to independent evaluation, we have no idea if they work at all or how well (or how poorly) they work.

If they were a medical product, the regulators would shut them down, but because martech is the wild West where anyone can make any claim they want without scrutiny, then we simply don't know well ANY of these tools actually perform other than the clearly overblown claims spilling out of their marketing departments.

Such independent validation is trivially easy to do...it's simply NEVER done!

Dale W. Harrison

6 of 44

7 of 44

8 of 44

9 of 44

10 of 44

11 of 44

12 of 44

13 of 44

?

14 of 44

15 of 44

16 of 44

17 of 44

18 of 44

19 of 44

Mi aspettavo una meraviglia, invece è stato una delusione!

Mi aspettavo una delusione, invece è stato una meraviglia!

20 of 44

21 of 44

LINK approfondimento

22 of 44

In ordine di priorità, dove si posizionano i metadati?

Links

Content

Html Template

Crawlability

URL Status code

Hosting & Database Uptime

Domain & DNS

CTR

Metadati

Schema.org is one of countless vocabularies (resp. ontologies). The DBpedia Ontology is another one. Both vocabularies are independent of each other. Another vocabulary would be The Organization Ontology.