CLOSE LOOK
Artificial intelligence: what’s the big story?

Artificial Intelligence, or AI, has long been hailed as the most significant technological leap since the Industrial Revolution. After decades in development, AI could now finally be fulfilling its promise. At best it’s viewed as a form of global digitalisation, allowing universal access to efficiency improvements, by automating processes and reducing costs. At worst it’s considered a sinister tool, that could develop a mind of its own. It has certainly been creating a media storm.

In essence, AI is software capable of solving difficult problems. It’s this constantly evolving software that allows machines to display a level of ‘intelligence’. The machines, or ‘bots’ are skilled at tasks such as search and logic. They are used in the fields of automation, sales, the law, translation, as ‘chatbots’ or as digital assistants. As with the steam-powered machines of the Industrial Revolution, they are designed to improve processes. But in just the same way, they could render certain human skills redundant.

The capacity of some chatbots is huge. Amazon-backed Alexa has 80,000 ‘skills’ and is the most financially successful chatbot to date. But the recent headline grabbers are the digital ‘friends’ or entertainers, such as ChatGPT. This chatbot uses ‘generative’ AI to respond to questions or requests, scouring the internet for the most probable answer. Critics have noted that these ‘probabilistic’ responses cannot always be relied upon. Nonetheless, this has been hailed as the true beginning of the AI revolution, prompting Microsoft to invested $10 billion in OpenAI, the developer of GPT-3 software.

But do recent advances in AI technology truly offer a global opportunity for progress? Regulators in China have banned access to ChatGPT’s uncensored content. However, the domestic internet giant Baidu is on the brink of launching its own chatbot, known as Ernie. Progress in China has been held back, as the range of internet data available for ‘training’ Ernie was constricted by past government censorship. And Chinese developers are currently hampered by US restrictions on exports of advanced semiconductors, making any training slower and more time-consuming.

And what of the costs? At a financial level, running ChatGPT alone is estimated to burn through $1 million per day. At a human level, the more vulnerable risk being duped by their digital ‘friends’, prompting EU legislation to ban passing off a chatbot as human. There are concerns that the days of ‘honest’ academic content are a thing of the past, while the new generation of software aimed at detecting AI-generated text has achieved only partial success.  But the hope remains that, once the hype fades, artificial intelligence will continue to optimise processes, bearing down on input costs as well as reducing energy usage in the march towards net zero.