How do you know what's real and what's AI? [Q&A]
The world is increasingly buzzing with stories about AI generated content. Today when you contact a business you might equally well be talking to a bot as a human, but is it possible to spot AI generated content and should we be worried about it?
We spoke to Gaurav Kachhawa, chief product officer of conversational messaging platform Gupshup, to find out how to distinguish AI-generated content, as well as the ethics that surround its use.
BN: How easy is it to distinguish between AI generated content and reality, and what are some tips for doing so?
GK: Content authentication is among the biggest challenges of our times. While there is enough research happening at companies like Meta, Google and some independent researchers as well to come up with a technical fix for the problem, given the pace at which GPT technology is evolving, tools to detect AI-generated content may get outdated very quickly. Having said that, there are ways in which it could be addressed. One way to solve it could be by providing source attribution. So, every AI generated content could come with a list of sources it referred to while building the response. The second thing which some companies such as OpenAI are trying to do is build plugins -- the idea being to limit the point of reference of an AI to only a certain set of credible sources. There have also been eager calls for bringing watermarks in AI generated content or for content creators to electronically sign their own created content. And lastly, one could use basic human logic and some pattern identification skills to differentiate an AI content from non-AI. For example, AI generated content will typically be very standard, coherent, lacking a human touch and tends to repeat certain kinds of words. All said and done, the space is rapidly evolving. It will require industry players to collaborate with LLMs.
BN: Is it true to say that tools like ChatGPT aren't 'intelligent' they’re just good at finding and presenting information, without understanding it?
GK: The term 'Artificial Intelligence' itself means intelligence that's artificially created, or created by humans. Therefore ChatGPT which is an AI model can't be intelligent in the true sense of the word. However, it's highly capable of ingesting vast amount of training data and using them to frame a response. When ChatGPT responds to prompts, there is no real 'thinking' happening. It also doesn't understand the text it so brilliantly spews forth. This is a mechanical procedure that groups and arranges words in accordance with an algorithm created by humans. That's why there are limitations to what ChatGPT can do. It can generate text-based responses to inputs, but it doesn't have the ability to generate creative content, solve complex problems, or make decisions that require human judgment. In cases where it can't come up with a response, it hallucinates or makes up an answer for the sake of it. There have also been instances where the model has not been able to understand sarcasm and humor.
BN: What are the ethical issues surrounding the use of AI and how can we tackle them?
GK: AI systems can show biases due to the data they are trained on, leading to discrimination. There are techniques for removing human bias -- when you train AI on the last 1000 years of human documents, it's going to be very biased. The text is very biased. But one needs to explicitly remove gender bias, racial bias and many other biases from that. The higher level issue is we just need to make sure that more of the good people have it than the bad. What gives me hope is with Open AI or some of these other Silicon Valley companies is the people creating it themselves are the most conscious of its power and its capabilities. And there’s already some governance mechanisms in place. There's a lot of conversations around it. At some point I'm sure the governments will have to get involved as well so that it stays in the right hands.
BN: How can conversational AI be used to improve the customer experience?
GK: Conversational AI is a transformative technology that offers a compelling ROI, while simultaneously improving customer experience. Driven by AI capability advancements that enable smarter, more scalable, and specialized solutions -- conversational AI tech, while being transformative, is also rapidly growing and evolving.
Across industry sectors, conversational AI solutions are helping several businesses drive more tailored consumer experiences. For example, in the BFSI space, AI powered conversational chatbots are redefining customer experience. Some of our customers are delivering hyper-personalized conversational banking experiences to their consumers. Use cases include automating routine tasks, providing quick responses to FAQs, enabling easy KYC updates and reducing customer waiting time for document processing, payment reminder messages with a one-click bill pay facility, easy access to available balance and other such account related information.
During the Covid-19 pandemic, when customers could not easily go to the market, retail brands world over were looking to build a Direct-to-Consumer channel and go-to-customer strategy. Especially in mobile-first countries such as India, APAC, LatAm -- the adoption of B2C interactions on social messaging channels have been a great booster of success of initiating conversational AI chatbot projects for retail brands. Globally as well, in a post-pandemic era, the lines between physical location, e-commerce, conversational commerce, and other forms of digital commerce are continuing to blur. Brick-and-mortar locations will need to pivot more to experience building centers to showcase what you can do with products versus just having a showroom of inventory to buy from. Commerce will become even more contextual with in-moment selling strategies on native channels of customer’s choice.
For example, we worked with a leading Austrian Skinwear brand to elevate shopping experience across customer touch points. With Gupshup’s retail AI brain-an enriched self-learning database, the brand saw a three percent increase in its quarterly online sales revenue. In yet another example, a global herbal wellness brand enhanced digital customer experience and automated 70 percent of customer queries with Gupshup's Conversational AI.
BN: Should we be worried about AI displacing humans in the workplace?
GK: On the face of it, it may seem that AI with human-like capabilities to write, draw art, code and fix bugs will be a replacement for many of these jobs. While that could be partly true as anything that's archetyped will likely get automated, a better way to look at generative technologies is considering them an auxiliary tool for productivity enhancement. Take for example generative art. When you have human designers working with AI, the prototypes can be faster. Smart journalists working with AI will be able to get more articles in less time.
In essence, it will be the human creativity and strategic thinking that will be at a premium. There will always be great value for understanding people and emotions, something no AI can do. Moreover, all of these technologies will need people to finetune them, optimize them and work with them. The skill set requirement may be slightly different. But that's where adaptation comes in with us standing testimony to Darwin’s theory.
It's a brave new world, no doubt, and those who adapt and embrace the change brought out by technology, end up gaining from it. So, rather than looking at it with skepticism or rejecting it, a good idea would be to upskill/reskill, focus on the creative parts and embrace the change. There is a larger good for sure in the technologies that have come up. After all, it's making a lot of the work better in quality, faster and efficient, but looking at it just with the prism of decimator of jobs would be undermining its potential.
As a society we will have to find the balance between the overall common good of AI and supporting and enabling people who are struggling with it.