What just happened?
I don't mean today, this week or this month. I mean what happened to AI industry in the last few months. There are thousands of articles on what jobs will be lost due to ChatGPT, but I couldn't find articles on the impact on traditional AI companies. How many will survive?
Let's take
Conversational AI, which is one area where plenty of AI companies had invested deeply. In fact, India was in the forefront of this area because of the need for supporting so many Indian languages for various business applications related to customer interaction.
All my views in this post are restricted to conversational AI and not other kind of AI.
Quote:
Originally Posted by https://www.ibm.com/topics/conversational-ai Conversational artificial intelligence (AI) refers to technologies, like chatbots or virtual agents, which users can talk to. They use large volumes of data, machine learning, and natural language processing to help imitate human interactions, recognizing speech and text inputs and translating their meanings across various languages. |
These are transactional bots, with specific knowledge of the customer data, in order to interact with the customer within a very specific context. In simple words, a transactional bot will glean the intent of the customer input (text or voice) and then reply with an appropriate response based on customer data. If you ask about weather or mutual fund products with a customer service bot, it wouldn't have a clue. Plenty of funded AI companies were working on building better and improved AI models to deal with more and more complex business interactions.
There are generally two kinds of business model followed by these AI companies. The first kind focus purely on developing technology using VC funding, and revenue generation is not a priority for them. They want to sell the technology or company to the highest bidder and then exit with their rewards. The second kind do focus a lot on revenue generation and use their in-house technology as a differentiator to provide more value to the customers. You could call them AI SaaS companies.
About 6 months ago, ChatGPT happens. A LLM AI engine that blows every in-house AI technology out of the water. It can instantly learn from any structured or unstructured data and generate very impressive (even if flawed) results. When it is exposed to customer data, it can conduct 10 to 100 times better interaction with the customer. The first kind of AI companies suddenly found themselves paddling a boat in a dry creek. Their base technology is nearly worthless, unless they can find some other kind of non-conversational usage.
The upcoming ChatGPT Business or a private instance of ChatGPT can wipe out AI SaaS companies too unless they quickly replace their in-house engine with ChatGPT APIs. This turns them from a product company to a pure services company. Their bottom-line will suffer, and they have to focus a lot on customer acquisition and develop very domain specific knowledge base.
BTW, a ChatGPT based transaction bot can actually discuss weather or answer unrelated queries during any business transaction. It will be an insanely knowledgeable agent or bot, on all matters.
So, what next?
Data analysts and data engineers will be still required to find and feed relevant data to ChatGPT. But what is the role of data scientists in the presence of ChatGPT?
And the biggest threat will be from traditional services companies like TCS/Infosys/Wipro/HCL and their ilk. With ChatGPT APIs, these service companies can easily out compete smaller AI companies, whose main USP just got neutralized. The large service companies have the connections and brand value to corner most of the business in conversation AI space thanks to ChatGPT.
In my long career, I have never seen a bleeding edge hi-tech industry get upended in this manner so quickly. I am still wondering if my analysis is even correct. Am I missing something here?
Can home grown AI industry push back against ChatGPT? Sam Altman recently commented on it, the answer is NO.
Visiting India, OpenAI CEO Sam Altman tells Indians: You can try to build AI like ChatGPT, but you will fail
And I agree with him. It is not a matter of skill or grit. It is a matter of funding and long gestation period. A ChatGPT beating technology will need few billion dollars of investment, and few years of waiting. Who will fund that in India? The risk of failure is simply too high.