Should we automate jobs? Is there merit to developing non-human minds that may eventually outsmart, outnumber, outdo (and even replace) us?
These issues were raised in an open letter by the Future of Life Institute, an NGO that called for a six-month pause in the development of artificial intelligence (AI). The open letter was signed by tech luminaries that included Elon Musk. This concern shows just how rapid progress is being made in AI that even VIPs are anxious about the potential dangers.
Large Language Models (or LLMS), the type that powers ChatGPT (a chatbot made by Open AI, a startup), have flabbergasted even their creators with their unexpected abilities. Their “emergent” abilities include everything from writing computer code to solving logic puzzles to identifying films from plot summaries written in emojis.
These abilities prove that we aren’t too far from the futuristic movie Ex Machina. The developments we are seeing are set to transform human relationships with computers. Proponents of AI argue that its potential to solve mammoth problems (such as developing new drugs, or designing new materials to help fight climate change) is too beneficial to stop AI’s momentum. But to many, the fact that AI’s capabilities are going beyond creators’ understanding brings to life the movie horrors of machines outsmarting their creators.
A decade ago, first-generation AI systems relied on carefully labeled training data. Once exposed to labeled examples, they can learn to transcribe speech or even recognize images. The systems of today no longer require pre-labeling. As a result, they can now be trained to use larger data sets culled from online sources. Simply put, LLMs can now be trained on the entire internet.
This capability became clearer with ChatGPT’s debut last November 2022. In its first week, a million were using it. In two months, it registered 100 million. It proved very useful for school essays and wedding speeches that its popularity prompted other rival firms to release their own versions.
Here’s the creepy part. When Microsoft incorporated ChatGPT into Bing, Bing Chat suggested to a journalist that he should leave his wife. ChatGPT has also been accused of defamation by a law professor. But the rationale is, LLMS produce answers that have some semblance of truth even if they may contain factual errors or even fabrications.
Hence, the call to “pause”.
The fear that machines will steal jobs is a century-old concern. LLMS are now doing skilled work and some white-collar tasks such as writing code or summarizing documents. But a closer look would show that technology has also spurred new jobs. There is still a gamut of tasks machines cannot do.
Nonetheless, the doomsday scenarios shouldn’t be dismissed. In the future, AI could potentially wreak havoc by having unrestricted access to money, energy, and computers. What we could do is regulate. The existing AI technology raises real concerns about privacy, bias, and intellectual property rights. The key is to balance the potentials of AI with an assessment of the risks AND once this is being done, to be ready to adapt. Most important, never underestimate the power of the human mind and the human heart. Nothing can replace the passion of real people at the other end. AI can systematize, but only a living, breathing person can empathize and deliver.
Talk to us.
NEARSOL is a US-based BPO and service company that offers clients custom-design solutions. With major hubs in Manila and Iloilo, it began operations in 2011 and has since gained a strong presence in the Latin American regions, the Caribbean, and the Asia Pacific, winning many awards for quality and service along the way. Most recent of which is its Great Place to Work-Certified™ Recognition, a global accreditation that tells stakeholders what employees think of the company culture. Visit our LinkedIn to learn more about our company and our global footprint.
