AI making inroads into the business world

SOFTWARE engineers are panicking that they may lose their jobs to the artificial intelligence (AI) software which can mimic a whole lot of things that they do to earn a living. And since India supplies the technology world with a massive number of software engineers — a boon for the educated middle class — what India is facing is a possible social turmoil resulting from the loss of large numbers of jobs for the middle class.

That this is not unnecessary panic too early in the day can be gauged from the fact that IBM is expected to pause hiring for roles that it thinks can be replaced with AI software. A survey has confirmed that AI has already begun to replace humans in the US. Perhaps the last straw is the ‘godfather’ of AI, Geoffrey Hinton, becoming so concerned about the risks that AI poses that he is leaving Google so that he can speak freely about these risks.

AI is a software that trains machines to perform intelligent tasks. Earlier, it was used mainly to decipher patterns and make predictions by analysing large amounts of data. But today the rage is generative AI, which can create words, sounds, images and videos like an intelligent human. It powers chatbots which ‘chat’, so to speak, with humans by answering questions instantly in eerily lifelike conversations.

The chatbot ChatGPT (GPT stands for generative pre-trained transformer) has taken the world by storm in no more than a few months. It can create photos and voices that have fooled experts and write software code, TV episodes, songs and even entire novels, all by mimicking humans with startling accuracy. It works through machine learning via neural networks that mimic the way the human brain works. All this it does by going through enormous amounts of data which it has to be fed.

Finding what ChatGPT, created by the San Francisco startup OpenAI, majorly funded by Microsoft, can do, other tech giants have got into the act. Google has come up with its chatbot Bard to power its search engine and Microsoft has unveiled Bing, entirely its own creation. These have created a boom in generative AI which represents enormous opportunity but, as can be guessed, is also capable of doing enormous damage.

The spectre that society sees before it is of deepfakes, hallucination and sheer misinformation. Chatbots have fathered imagined scandals, conspiracy theories and plain racist texts. These would be grist to the mill for troublemakers such as the Donald Trump supporters who had stormed the US Capitol on January 6, 2021, seeking to thwart the certification of Joe Biden’s victory in the presidential election.

Before we start panicking, there is a need to also look at the other side of the picture. First, generative AI does not come cheap. OpenAI, which has developed ChatGPT, lost a massive $450 million last year. It is likely to turn out to be the most capital-intensive startup in the history of Silicon Valley. But the market wants the product. OpenAI is likely to clock a revenue of $1 billion next year.

Second, generative AI will also affect the job market in a positive way. There is a serious shortage of data science specialists working on AI and some of them are getting a pay hike of 35-50 per cent when they are switching jobs. Companies in as diverse fields as healthcare, finance and entertainment are seeking to build their own AI engines. The demand is coming not from software vendors to whom routine work is outsourced — the backbone of Indian IT — but the companies or clients themselves as AI work cannot be outsourced.

Why do companies need AI capabilities? They help respond to messages, generate ideas, navigate tough communications, facilitate team conversations, and help write self-assessment reports. They perform routine tasks like writing emails, other than, of course, writing code. However, the text generated has errors, sometimes contradicting itself, and the code written has bugs. Overall, AI chatbots provide a great starting point, but the output has to be overseen to correct errors. Simply put, it is a great help to get a first draft of a response to an email but you cannot let the mail go unsupervised.

The Indian software scene faces contending realities. AI can do the absolute basic work of writing a code or routine tasks of system maintenance. But the output has to be checked for errors. Over time, as AI systems improve, there will be more and more self-correction, but the systems themselves will have to be taken forward. So, the Indian software industry will not need the number of engineers that it currently does but the demand for output and systems supervisors will go up. Fewer low-paying jobs but more high-paying jobs will be the scenario of the future. And that future can arrive within a year.

At the end of the day, why will companies which have nothing to do with software directly, say a steel or beverage company, need to keep abreast of developments in AI? A key reason is that the field is so new, the future is so open and the potential so vast that no company worth its salt will want to be left out of the mainstream of developments in AI.

It is like asking if a company needs a powerful IT system. Of course, it does. AI is really like the next layer. It uses IT to do the routine computational work that any business has to do, either in-house or outsourced. The challenge that AI offers is getting the IT to do a part of the routine cerebral work too so that the number of staff in non-customer facing positions is reduced. Other than writing or vetting resumes, what AI can do is write the first draft of the company’s annual report.

And should those in the government think that generative AI is a headache for the private sector, the realisation should dawn that the jobs of those who do the first draft of notes are also at stake. Generative AI can even do a first take on the Budget speech of the Union Finance Minister!