LUMEN | ARTIFICIAL INTELLIGENCE - A BOON OR A BANE ?
Besides the ‘ deep learning ’ of our online user habits , AI also has brought about some positive impact in society , such as relieving the workforce gaps in many industries such as customer service . AI chatbots are prevalent on many websites and online applications . Based on a study by Abby McCain for Zippia . com , chatbots were projected to have handled 90 % of customer interaction without human input in 2022 .
Yet , there is a growing nagging concern about the negative impact of AI that is in need of rectification . The rise of misinformation during COVID has been aggravated by the in-tandem rise in AI usage . Regulators around the world are scrambling to adapt existing rules and create new guidelines to govern the use of generative AI that can create text and images and at the same time , engendering excitement and fear about its potential that can reshape and disrupt many industries . In Singapore , tech giants and major global industry players in AI have allied with the Infocomm Media Development Authority ( IMDA ) to tackle pressing issues in AI such as bias , copyright and its susceptibility to lying . The Singapore-based AI Verify Foundation which was launched in early June 2023 will discuss AI standards and best practices , and create a neutral platform for collaboration on governing AI .
Coming back to the most enlightening ChatGPT , the AI model does have some limitations in the data it relies on when answering our questions . For example , ChatGPT has a data cut-off where its Internet information base does not go beyond 2021 . The ideal answer it churns out depends on what the model knows , rather than what the human demonstrator knows .
ChatGPT also surprisingly makes errors in some simple mathematical questions . I successfully tricked ChatGPT into thinking that 30 + 28 is 57 just by pointing out its “ mistake ”. This could mean that ChatGPT will consider my answer to be valid , which can lead to this data being placed into its database . These errors may seem small but can lead to widespread misinformation generated from ChatGPT . Another way AI is used for misinformation is through deep fakes where AI is used to create fake videos and pictures using pre-existing or generated media . For example , art made with AI-generated content has been winning art competitions . An AI-generated work titled “ Théâtre D ’ opéra Spatial ” ( sounds
fancy ) won an award in the digital category at the Colorado State Fair . The artist , Jason Allen , simply had to type a string of words into an AI image generator called Midjourney and an awardwinning art piece was created . This raised many questions in the art community on whether such forms of art should be considered a deserving piece of creativity or if it is just plain cheating .
Another example is the bizarre doctored photo of Pope Francis in a fashionable Balenciaga puffer jacket that went viral on social media . While some minor unrealistic flaws were spotted , the picture looked largely believably real ! As AI inevitably develops further , it won ’ t be long until a real picture is indistinguishable from a deep fake .
AI might seem like an abstract concept to many , yet its use is actually already widespread in almost every aspect - from real to unreal . The next time you use ChatGPT for your homework assignment , think twice if the information is valid and credible enough to be used . AI may not always be right .
As ChatGPT once said : Why was the robot excited about going to school ?
Because it heard there would be a byte-sized lunch !
ChatGPT doesn ’ t have a sense of humour ( yet ).
1
Source : https :// www . ibm . com / topics / artificial-intelligence ,
2
Source : https :// www . sas . com / en _ sg / insights / articles / analytics / five-ai-technologies . html
3
Source : https :// openai . com / blog / chatgpt
18