DIGITAL INTERACTIVE SERVICES

How to spot AI-generated content

Is this article written by a human or generated by Artificial intelligence (AI)? Rest assured, this one is human-generated! However, this thought may dwell in people’s minds while looking at future digital content. Recently ChatGPT from OpenAI, the latest generative AI tool that can mimic human responses to queries, has taken the world by storm. A refined conversation with ChatGPT can replace hours of sifting through search results to find structured information! It can even write software code. It appears the buzz will only intensify with Google announcing the launch of Bard AI and Microsoft unveiling a new version of their search engine Bing, incorporating the ChatGPT technology. 

What can AI-powered chat platforms do? From a business perspective, there is potential for customer service solutions to be revolutionised in a big way, and so are digital marketing services. These chatbots will drive enhanced customer experiences while customers are shopping or interacting with customer support. Businesses can use them at the backend too. They could converse with them as they do with digital marketing consultants to generate content for marketing collaterals. On a personal level, individuals can chat with them for inputs on any topic and even get poems or articles written. They can access far more sources than humans can and in much lesser time, to generate content. So, does it spell doomsday for human content creation, or do we still have a few aces up our sleeves?

Thankfully, these systems have some catching up to do to draw level with the quality of original human writing. To know if what we are reading is AI-generated, we can watch out for some telltale signs. There are always chances of some hits and misses. Not all AI-generated content will have issues, and some human-generated content may have glitches similar to AI-generated ones. However, at least for now, a good blend of some tips and a few technical tools can considerably help spot AI-generated content. AI models are catching up fast, and we must keep evolving faster. Here are six tips you can use to try and spot AI-generated content.

No unique perspective:

AI-generated articles are highly generic and present facts and figures flawlessly however, no distinctive author viewpoints based on their knowledge or experience come forth. A distinct lack of perspective is sure to be felt as as the model doesn’t have opinions and only presents facts. There is no specific writing style, as AI has no personality and its output is based on an algorithm. Also, most writers finetune the content for the intended audience with a story-telling approach. This aspect cannot be replicated yet in AI-generated text.

No lengthy sentences:

AI-generated content provides clear and concise information using short sentences. However, in human writing, there are chances of sentences ranging to monstrous proportions in trying to explain a concept, that they forget the objective of presenting the information crisply! Oops! That was lengthy, huh? Yes, in human-generated content, you would almost certainly find a few sentences that do not fit the concise bill.

Repetition of words and phrases:

AI tries to fill gaps with relevant SEO keywords or fluff. You can spot the overuse of a few words such as ‘it’ and ‘is’. AI-generated content follows an algorithmic pattern to display unbelievable consistency and predictability. As they say, to err is human, and human-generated content will have some slip-ups somewhere. Also, unlike humans, AI is likely to almost always use commonly used words.

Inaccuracy:

Accuracy is supposed to be the hallmark of AI. So, can there be inaccuracy in AI-generated content? AI is good at collating data from many sources, analysing and providing insights based on patterns. However, if the AI algorithm encounters conflicting data from multiple sources or accesses inaccurate sources, it may get confused. Whether predicting numbers or dishing out product descriptions, AI can trip up, and a trained eye can spot the discrepancies easily.

Missing sources:

 Check if the articles quote any sources and how legit they are. AI-generated text will not have source links based on which it gave you the inputs. Hence, articles with no source credits or mentioning questionable sources are signs of AI-generated content.

Use tools:

There are tools available to detect AI-generated content with varying degrees of accuracy. OpenAI Classifier from OpenAI can help determine if the content is AI-generated. However, the efficacy is limited as there could be false positives. Another tool, Originality.ai, helps identify AI-generated content and plagiarism. After the recent ChatGPT buzz, they published their tool accuracy levels on various generative tools on their website.

In conclusion, rapid development is happening in the AI domain, leading to generative models becoming more intelligent. However, lack of trust due to missing sources for the output is a drawback. A good combination of tools and human judgement is the best way to spot AI-generated text. The detection tools must match the speed to stay ahead. Only time will tell if human intelligence can beat AI and stay ahead.  

*For organizations on the digital transformation journey, agility is key in responding to a rapidly changing technology and business landscape. Now more than ever, it is crucial to deliver and exceed on organizational expectations with a robust digital mindset backed by innovation. Enabling businesses to sense, learn, respond, and evolve like a living organism, will be imperative for business excellence going forward. A comprehensive, yet modular suite of services is doing exactly that. Equipping organizations with intuitive decision-making automatically at scale, actionable insights based on real-time solutions, anytime/anywhere experience, and in-depth data visibility across functions leading to hyper-productivity, Live Enterprise is building connected organizations that are innovating collaboratively for the future.