The $100 Billion Hallucination
Artificial intelligence has been the buzzword for some years now. Every industry is looking to adopt it in some manner to reap the benefits. It is aimed a simplifying our lives and the arrival of ChatGPT has only reinforced the belief in the simplification AI (Artificial intelligence) will bring in a human’s life. Last week witnessed 2 tech giants, Microsoft and Google face off in the artificial intelligence space with google announcing it experimental conversational AI which was meant to be better than ChatGPT and Microsoft announcing New Bing powered by Open AI. Referring to the previous article on ChatGPT (https://medium.com/@ohri1294/ok-google-is-it-your-doomsday-dfc121e07106), this was something bound to happen with the technological behemoths realizing that they were falling behind in the face of experimental conversational AI.
Despite being a witness to 2 big announcements within the week, the most eye-catching headline in the world of technology and business was Google losing almost 8% of its share value, i.e., $100 Bn in market value owing to factual error displayed by its experimental AI engine named BARD by indicating inaccurate information related to James Webb Space Telescope. Such a factual error indicated that cracking artificial intelligence is not easy, and everyone may be hallucinating in the name of competition to bring this for general availability to the public. This could not only be a $100 bn hallucination but a hallucination beyond imagination which every company, especially Microsoft and Google must be cognizant of while adopting Conversational AI. There are many angles to the hallucination which can be outlined below:
1. Machine Leaning is not high accurate: Let’s face it that machine learning and AI are learning information based on data available and can often learn incorrectly owing to noise in the data or discrepancies in the pattern. This makes the outputs of AI based engines high susceptible to inaccuracies leading to increased distrust among the public.
2. Market Shares are at stake: Companies like Google command 84% market share in search while Microsoft is looking to diversify its market share from Cloud services and Desktop business. The $100Bn fiasco this week indicates that rushing the release of a not very well refined product can lead to market share loss in this cutthroat industry and gives an opportunity for even the small companies to make a mark. This will again impact the belief system of the public in Googles and Microsofts. Not to forget, the extensive layoffs that can happen again.
3. People still believe in search: While ChatGPT has been a rage for the past few months, people still believe in search with google search being the first resort for any information that people seek. Modifying its essence with inaccuracies will only degrade the belief system in search. The same can be said for Bing which cannot run the risk of disowning its fundamentals in the face of stiff competition.
4. Never Ending Testing: ChatGPT has simplified lives, but it is still prone to errors and requires extensive testing. It is a good idea to be in an experimental state with extensive testing instead of having general availability of the new features of Bing or BARD to avoid backlash on the technology and allowing ChatGPT to take center stage again.
The aura of conversational AI is here to stay, but the path to be treaded must be very careful as any miss from perfection or any instance of hallucination to capture the market share or public’s attention without detailed testing will lead to severe repercussions which will lead to a hit on the reputation of the companies and unravel the beliefs of the public. The way forward is to continue testing and refining the deep learning models and release mini versions to the public which are high accurate to build the confidence before launching a wholesome product/feature. This would lead to not a drop of $100bn but a rapid increase in market value in the magnitude of $100 bn.
