By Jaspreet Bindra
If late 2022 was all about the jaw dropping experience that OpenAI’s just launched ChatGPT brought to us, late 2023 was dominated by the deepfakes we encountered and countries bringing out AI Safety principles and regulations almost on a daily basis. In a year, the narrative changed from unbridled excitement to a nagging concern on what this shape-shifting technology and do our lives, our jobs, our businesses and our planet.
2024 promises to be the AI Year, where businesses will discover new use cases, tech companies will further embrace AI and GenAI reshaping the tech landscape, and humanity will try and cope with the ethical issues AI throws up. To me, three big issues will dominate the moral and ethical questions about AI.
Deepfakes and the threat to democracy: 2024 is the year of elections as the largest democracies around the world – India, US, UK, Indonesia, etc. – undergo this important ritual of democracy. What happens in these elections, especially in the US and India, is going to hugely influence geopolitics, climate impact and the economy of the world. There is an enduring fear that deepfakes will be used to sway the narrative and manipulate voters minds through creating false but authentic-looking videos and images of political leaders. The Indian Prime Minister declared in a speech recently that he himself had become a victim of a ‘deep fake’ being shown to having participated in a folk dance, which he did not. Donald Trump was shown to be arrested by the Metropolitan Police in the US. Another was of a debonair looking Pope in a white Balenciaga puffer jacket. While countries are racing to build regulation around deepfakes and AI and social media companies claim to be doing their best to address this issue, this will be one of the key ethical quandaries of 2024
The copyright and plagiarism debate : The US screenwriters went on strike last year to protest against AI generated work, many famous authors launched a class action suit against AI companies, and in the last week of December 2023, the New York Times sued Open AI, Microsoft, and Google for copyright infringement of its content for designing their Large Language Models. This issue is fraught and a legal minefield, but the outcome of these cases will decided the future and direction of Generative AI. This is an old ethical quandary – who owns the content and how is is protected, but Generative AI has brought it to the forefront.
The threat of environmental degradation by AI: this is not something which is in the centre of the debate now, but I suspect that it will get more attention as the world warms up alarmingly. LLMs can be notorious energy hogs; for instance, training a large transformer model just once would have the equivalent CO2 emissions of 125 NY-Beijing round trips. And this is for just training it once for a model with just 213mn parameters; GPT3 has 175bn and the incoming GPT4 has a rumoured 100trn! About 2% of the world’s CO2 emissions are attributable to the ‘cloud’, and these humongous models are growing the cloud, and therefore the emissions, in a quantum way. There is also the environmental impact of the powerful chips that power them, which use millions of gallons of pure water, and massive power.
The good news is that the world is waking up to the ethical issues around AI. It is super powerful technology which can do great good, but like nuclear energy, also has a dark side. Unlike in social media, for example, which was left largely unregulated, AI regulation and safety is being taken up at the highest levels by both countries and corporations, and we hope that 2024 will see humanity creating the right guardrails to manage these ethical dilemmas.
The author is managing director and founder, The Tech Whisperer
