What is the quality of the data used to feed the AI content generators?

Artificial Intelligence (AI) content generators are becoming more widely talked about, so Lacewing Media have been giving thought to how they can (and no doubt will eventually) work alongside professional marketing services. We want to question the accuracy of data used by these content bots – is it robust enough? For these content generators to do their job correctly, it is essential that the data feeding into them is accurate and unbiased, otherwise the content itself will be inaccurate or irrelevant, ultimately making the tool less useful or relevant to marketeers.


How accurate and useful is the content generated by AI?

Language processing AI models, such as Chat GPT, generate human-like content. They are trained to generate content using large text databases from the internet such as books, webtexts, Wikipedia, articles, and other pieces of writing on the internet. These models work by entering an instruction (prompt) to generate content such as articles, blogs, emails and social media posts. Considering the data sources these models are trained on there is a question about its robustness and accuracy. These models have access to internet forums and articles, so access to detrimental data can be high, especially in a regulated industry such as the alternative investments sector.

The Data Scientist claims “after having carefully reviewed the arguments of the various automatic writing sites, nothing is said about the way in which these tools recover information. There is therefore no guarantee that the sources used are verified.” Open AI openly admits that one of its models Chat GPT “sometimes writes plausible-sounding but incorrect or nonsensical answers” which is highly concerning.


Could AI generated content be offensive?

Because content generators such as GPT-3 and Chat GPT take information from the internet it can generate content based on bias commentary, stereotypes or general opinions therefore it could potentially generate offensive or out of date content. Open AI state that “while we’ve made efforts to make the model refuse inappropriate requests, it will sometimes respond to harmful instructions or exhibit biased behaviour. We’re using the Moderation API to warn or block certain types of unsafe content, but we expect it to have some false negatives and positives for now.”


Are AI content generators suitable for the alternative investment sector?

The alternative investment sector is a highly regulated industry with a constantly evolving landscape, therefore the importance of accurate and reputable data sources is essential. Thought leadership is an important marketing strategy for investment firms as content is written to innovate and challenge trending topics. This process ensures industry leaders are keeping on top of what’s happening in the market and provides confidence in clients and potentials that these firms are experts within their field and can be trusted with their investments.

AI content generators cannot be used to create innovative content based on complex subjects and opinions. At present it can only generate content using past data and cannot quote currently trending topics nor have an emotive opinion on them. Chat GPT, for example, is available for public testing and only uses data prior to 2021, so is already out of date for the alternative investments sector. Although it’s inevitable that new AI technologies are evolving and businesses should embrace them, it’s also advisable to remain cautious when using them. It’s our responsibility as marketeers to use them in the most worthwhile way to maximise efficiencies within our business and our clients, but at present these kinds of tools would only restrict agile thought leadership in a market which has constantly evolving regulations.