Advancements in artificial intelligence

A

Have you tried OpenAI’s ChatGPT lately? OpenAI, an artificial intelligence (AI) research company originally founded by Sam Altman amongst others, have the mission to develop AI tools to, as stated on their website, “ensure that artificial general intelligence benefits all of humanity.” ChatGPT—which is an AI-powered chat robot—was launched by OpenAI in November 2022, and its success has led many tech companies to adopt similar technology in their own products. The difference between this chatbot and other similar chatbots that have come before it is that this one generates very good responses to questions in a way that mimics a real human conversation. In fact, according to ChatGPT itself, it is different because it is trained on a much larger and diverse dataset; it uses “deep learning algorithms” to adapt to new information; it has advanced natural language processing capabilities and contextual understanding. I tried it myself for the past couple of weeks, and needless to say…this thing is impressive. Recently, Microsoft sent out the first wave of ChatGPT Bing invites to use their chatbot. Microsoft, having invested billions of dollars in ChatGPT maker OpenAI, announced an extended collaboration and investment with the company. The invitations to try out the new version of the search engine have led to a lot of interesting user experiences: some have claimed to have been threatened, others claimed they were given wrong information to users while asserting that they were correct and declaring love for its users. According to Kevin Roose, a New York Times columnist, the bot urged him to leave his wife after declaring its love. His final thoughts about the bot? It left him feeling “deeply unsettled, even frightened.” He worries that it may influence its users to act in harmful ways and carry out dangerous acts. In my experience, while ChatGPT is amazingly impressive (and can give you information/ideas regarding problems that you did not think of), it certainly is not perfect. For example, a Forbes article (and multiple Reddit posts) found that ChatGPT has been referencing fake scientific abstracts/articles as sources for their information. Some of these sources don’t exist, yet it was able to fool even scientists that they were real reports. Regardless, it does not lessen the fact that ChatGPT is a significant advancement of AI: it is able to hold a human-like conversation and respond intelligently. It was able to also contextualize the conversation. For example, I told the AI a joke, the chatbot tried to understand the joke, and when I said that the point of the joke was missed, it apologized and asked me what the punchline actually was. Google also has plans to roll out an AI system, but they are aware that they are not ready yet. In a leaked company email, Business Insider reports that Google is asking its employees to take a few hours of their day to test out “Bard” AI, a chatbot that will compete with Open AI’s ChatGPT model. Bard is expected to have access to the data and to be integrated with Google’s search engine, which may be a huge advantage against Microsoft’s ChatGPT. Perhaps the way that we use search engines today will become a thing of the past once AI-assisted web searches are well adopted. Some have concerns about the impact of AI in today’s society (such as university academics warning that it could threaten university education, as students use it to complete essays), but there are certainly many beneficial ways that AI can be implemented: AI can make a huge difference in the implementation and development of healthcare, advancements in transportation, and can play a role in almost everything. It can write, converse, and even pass tests. Law professor Dan Schwarcz from the University of Minnesota stated that he could not tell the difference between exams written by an AI and a student, and a test taken by ChatGPT passed with a C+ (if you’re interested, look up “ChatGPT Goes to Law School”). ChatGPT has also passed the US Medical Licensing Examination (USMLE), without any specialized training. AI can play a huge role in other more substantial matters: On Monday, 14 February, the US Department of Defence’s research agency announced that AI algorithms can control an F-16 in flight. On 16 February 2023, the US launched the use of AI and autonomous weapons by militaries. However, they recognize that there are concerns about autonomous weapons and there should be responsible military use of AI. There are certainly valid concerns that come with the development of AI, but it is here to stay. It goes without saying that these recent advancements in artificial intelligence will certainly shape the future of humanity and its rapid evolution will continue to have tremendous impacts on how we live our lives.

About the author

Melannie Freza
By Melannie Freza

Monthly Web Archives