The arrival of new AI large language models has revolutionized the world. From altering businesses to making a change to consumer and institution-to-institution modes of communication, it’s everywhere.
A new study by Weixin Liang from Stanford University highlights the shift is serious and it portrays how LLM-assisted writing is on the rise in different sectors. We’ve got press releases from the corporate world and other types of job postings present to provide consumer grievances, not to mention the UN reports.
By the end of last year, we saw plenty of AI-based writing being at the forefront with nearly 24% of business press releases, 18% of financial grievances, and 14% of press releases from the UN entailing some kind of AI edits or writing services.
The authors shared how they curated a technique for quantifying and relating different frequencies of terms which are the most and less likely used by AI systems. After that, they studied the prevalence rates through various kinds of text.
The study is among some of the biggest investigations for adopting AI writing and it can review nearly 300M online papers and content between the last four years.
There was a time when AI-based content was limited to tech-savvy organizations and the usage went up after ChatGPT’s launch during the latter part of 2022. In just a few months’ time, we saw proof of how Generative AI content started to spread online.
Smaller companies were leading the way and AI-based jobs were soaring to nearly 15% at smaller organizations as per the study. Corporate earnings followed this similar trend and it surged to 24% during the latter part of 2023 before being leveled off.
The biggest takeaway here is how swiftly this LLM-based writing grew and how big organizations such as the UN are now using them in 14% of all press releases. This frequency indicates the potential and dangers that come with AI writing.
LLMs bring back efficiency as professionals can produce more content and get rid of confidence having to do with authenticity. The massive growth of these models in writing also gives rise to a new combo of pros and cons. Remember, AI can assist in enabling speakers who aren’t native to better express themselves during a conversation and this gives rise to a new era of formal chats.
For instance, anyone claiming they wrote one of the biggest consumer financial complaints in a sample explained how chatbots helped in better organizing thoughts and comprehending rights and laws.
But the issue is that it comes with a major disadvantage in terms of authenticity and suspicion. The research highlights how it predicts dangers and makes it more difficult for firms to get things to stand out. Another major issue is job listings where the majority are based on AI at smaller organizations.
The research highlights the issues and template-based writing arising in different voices. It’s hard to find anything unique and with competition growing, that can be a serious issue. Imagine going for a job with the requirements generated by AI. Candidates end up struggling to get a chance to succeed at any real opening due to this.
See the issue is that the more AI-based content there is, the higher the chances that future LLMs get trained using synthetic material. This opens a new pandora of issues including disinformation, bias, and reduced creativity.
Thanks to research published in Nature magazine in 2023, AI models get trained on different types of content that are made using AI instead of diverse material authored by humans. So in essence, the AI models are feeding off their own reflections and producing content that’s so unreliable. Remember, this is not a trend, it’s a shift that’s altering the internet.
Image: DIW-Aigen
Read next: OpenAI Releases Exciting Updates for its Advanced Voice Mode Enabling Chats in Real-Time on ChatGPT
A new study by Weixin Liang from Stanford University highlights the shift is serious and it portrays how LLM-assisted writing is on the rise in different sectors. We’ve got press releases from the corporate world and other types of job postings present to provide consumer grievances, not to mention the UN reports.
By the end of last year, we saw plenty of AI-based writing being at the forefront with nearly 24% of business press releases, 18% of financial grievances, and 14% of press releases from the UN entailing some kind of AI edits or writing services.
The authors shared how they curated a technique for quantifying and relating different frequencies of terms which are the most and less likely used by AI systems. After that, they studied the prevalence rates through various kinds of text.
The study is among some of the biggest investigations for adopting AI writing and it can review nearly 300M online papers and content between the last four years.
There was a time when AI-based content was limited to tech-savvy organizations and the usage went up after ChatGPT’s launch during the latter part of 2022. In just a few months’ time, we saw proof of how Generative AI content started to spread online.
Smaller companies were leading the way and AI-based jobs were soaring to nearly 15% at smaller organizations as per the study. Corporate earnings followed this similar trend and it surged to 24% during the latter part of 2023 before being leveled off.
The biggest takeaway here is how swiftly this LLM-based writing grew and how big organizations such as the UN are now using them in 14% of all press releases. This frequency indicates the potential and dangers that come with AI writing.
LLMs bring back efficiency as professionals can produce more content and get rid of confidence having to do with authenticity. The massive growth of these models in writing also gives rise to a new combo of pros and cons. Remember, AI can assist in enabling speakers who aren’t native to better express themselves during a conversation and this gives rise to a new era of formal chats.
For instance, anyone claiming they wrote one of the biggest consumer financial complaints in a sample explained how chatbots helped in better organizing thoughts and comprehending rights and laws.
But the issue is that it comes with a major disadvantage in terms of authenticity and suspicion. The research highlights how it predicts dangers and makes it more difficult for firms to get things to stand out. Another major issue is job listings where the majority are based on AI at smaller organizations.
The research highlights the issues and template-based writing arising in different voices. It’s hard to find anything unique and with competition growing, that can be a serious issue. Imagine going for a job with the requirements generated by AI. Candidates end up struggling to get a chance to succeed at any real opening due to this.
See the issue is that the more AI-based content there is, the higher the chances that future LLMs get trained using synthetic material. This opens a new pandora of issues including disinformation, bias, and reduced creativity.
Thanks to research published in Nature magazine in 2023, AI models get trained on different types of content that are made using AI instead of diverse material authored by humans. So in essence, the AI models are feeding off their own reflections and producing content that’s so unreliable. Remember, this is not a trend, it’s a shift that’s altering the internet.
Read next: OpenAI Releases Exciting Updates for its Advanced Voice Mode Enabling Chats in Real-Time on ChatGPT
