The title Microsoft GPT-4 reveals the company’s latest efforts to develop artificial general intelligence (AGI). A recent paper on the arXiv preprint server entitled “Sparks of Artificial General Intelligence: Early tests with GPT-4” announced Microsoft’s claim that they have achieved early AGI capabilities with its AI. This claim is contested by Sam Altman, CEO of OpenAI, which Microsoft is heavily investing in to incorporate OpenAI’s GPT language models into their products.
The study generally follows the frameworks defined by a group of psychologists from 1994 of what AGI should be like. GPT-4’s capabilities are on par or even surpass humans in some aspects, such as its ability to draw a unicorn in TiKZ, produce a proof with rhymes on every line, and manage a dialogue in a chatbot. Although GPT-4 also has many flaws and biases like sensitivity to inputs, complicated long-term memory, and lack of transparency and consistency.
Microsoft, a company well-known for its Windows operating system, is a leading technology multinational that has become a major player in the digital sector and whose strategies focus on cloud computing and artificial intelligence. Microsoft’s CEO Satya Nadella’s vision is to empower every person and organization on the planet to achieve more. With its investment in OpenAI, Microsoft aims to bring its AI principles and cloud computing technology to industry-leading AI capabilities. meanwhile Sam Altman, CEO of OpenAI, has clearly made it imminent that GPT-4 is still far away from being an AI.
In conclusion, the article both illuminates and questions the promises Microsoft makes about its AI capabilities and for AGI in general. Although Microsoft’s AI may have generated some sparks, it is far from enough for it to be declared true AGI.
