Generative AI And Its Benefits And Blind Spots!

Over the last few years, Artificial Intelligence (AI), particularly Generative AI, has garnered a lot of attention from top tech companies and startups as well as from the public. Generative AI is a type of Artificial Intelligence that can create images, sounds, lyrics, text and more based on program training data and a set of inputs/requests. Essentially, such programs can generate results faster than humans.

One such tool is ChatGPT, which was launched in November 2022 and received massive praise amassing over a million followers in less than 5 days and over 100 million active users in 2 months. ChatGPT is a text-based Generative AI that answers users' questions. Upon release, experts and thought leaders praised this new technology saying it could impact many industries including media and healthcare. This was further evident as ChatGPT recently passed all three parts of the U.S. Medical Licensing Examination.

Additionally, ChatGPT’s creator OpenAI has received funding from Microsoft on several occasions, who invested $1 billion in 2019, again in 2021 and now in January 2023, reportedly invested $10 billion over a few years. This partnership enabled Microsoft to introduce the technology as a chatbot in its search engine Bing (available to a few people) and plan to use it in other products.

Speaking on the same, Satya Nadella (Chairman and CEO, Microsoft) said, “We formed our partnership with OpenAI around a shared ambition to responsibly advance cutting-edge AI research and democratize AI as a new technology platform.”
 
However, despite its capabilities in games, videos, images, lyrics, chatbot interactions and other expressive outputs, Generative AI can’t revolutionize every industry. Especially since it can end up being costly and even dangerous in mission-critical situations.

Users of Generative AI tools have already reported instances where the chatbot has provided incorrect answers, in addition to Bing’s chatbot producing incorrect answers, threatening responses and concerning behavior (read here). Similar issues were noted with Google’s Bard (read here). Additionally, according to some studies, an estimated 20% of answers provided by ChatGPT are made-up, which, over time could lead to a spread of misinformation.

Experts believe that under certain situations Generative AI might seem valuable but could lead to complications. Certain results produced by the tool may be correct but may not be achievable due to insufficient materials or limited resources. Since Generative AI models are considered “black box” models, which offer no clarity on their internal workings. This makes it difficult to determine facts and identify errors or biases in the produced results, especially considering a key source of ChatGPT’s training data is Reddit.

According to a study (read here) by Emily Bender, Timnit Gebru, Angelina McMillan-Major and Margaret Mitchell, “large datasets based on texts from the internet overrepresent hegemonic viewpoints and encode biases potentially damaging to marginalized populations.”
 
Since it's believed that Generative AI could produce tremendous business value, experts suggest people should remain excited but proceed with caution.

What do you think of the benefits and use of Generative AI and do you think its current form raises cause for concern? Let us know in the comments below!