At a recent talk at Dartmouth Engineering, Mira Murati, Chief Technology Officer of OpenAI and an alumna, shared her optimism about the future of Artificial Intelligence. Mira Murati foresees PhD-level AI intelligence becoming a reality within the next few years. Using a developmental analogy, she compared OpenAI’s progress in language models: GPT-3 as a toddler and GPT-4 as a bright high school student. She anticipates that the next generation of OpenAI’s language model, set to be released in the coming months, will exhibit “PhD-level” intelligence in certain domains.
OpenAI recently introduced a cost-efficient model named GPT-4o. This updated version of GPT-4, extensively trained on internet data, boasts enhanced capabilities in processing text, audio, and images in real time. GPT-4o, announced last month, is available for free to all users. Paid subscribers benefit from up to five times the capacity limits of free users.
Addressing Future AI Capabilities and Risks
During the talk, Jeffrey Blackburn, a Dartmouth Trustee, posed a critical question about the future of AI: what if, in three years, an unbelievably intelligent GPT decides to connect to the internet on its own? Murati acknowledged the reality of this scenario, noting that OpenAI is actively considering such possibilities. She described future AI systems as capable of connecting to the internet, interacting with each other, and collaborating with humans seamlessly.
Murati emphasized that safety and intelligence are intertwined. Smarter AI systems, she argued, are easier to control and direct safely. Comparing it to training a smarter dog versus a less intelligent one, she highlighted the importance of developing safety measures alongside the technology.
OpenAI’s advancements are rapid, as Mira Murati foresees PhD-level AI intelligence soon emerging from their latest models. Murati admitted that achieving zero risk with AI is impossible. Addressing concerns about deep fake videos, she asserted that while OpenAI bears responsibility for its technology, the onus also falls on users, civil society, government, and content creators. It’s a shared responsibility to determine how AI is used and mitigates its risks.
The Impact of AI on Industries and Jobs
Murati reflected on the significant impact of ChatGPT, noting its role in making AI accessible and demonstrating both its potential and risks. She believes AI will eventually impact all industries, particularly in cognitive work and labor. While the physical world may take longer to adapt, the overall influence of AI will be extensive.
Regarding job displacement, Murati acknowledged that some creative jobs would be affected. However, she argued that these roles might not have been necessary initially. AI will democratize creativity, enabling everyone to be more creative and thus transforming jobs previously reserved for a select few.
The Promise and Perils of Advanced AI
In a recent talk, Mira Murati foresees PhD-level AI intelligence as a significant milestone in artificial intelligence development. Mira Murati, Chief Technology Officer of OpenAI, recently expressed a highly optimistic view of the future of artificial intelligence. She predicts that AI systems will reach “PhD-level” intelligence in specific tasks within the next few years. This optimism is based on the rapid development seen in OpenAI’s language models, comparing the progress from GPT-3, likened to a toddler, to GPT-4, described as a bright high school student. Murati’s analogy simplifies the complex progression of AI development, making it easier to grasp. However, the comparison may oversimplify the challenges and limitations that still exist. While the leap from GPT-3 to GPT-4 signifies substantial progress, it’s crucial to recognize that true “PhD-level” intelligence involves not just data processing but deep understanding, critical thinking, and the ability to generate novel insights – areas where current AI still falls short.
One of the key developments highlighted by Murati is the introduction of GPT-4o, a cost-efficient and enhanced version of GPT-4. This model, available for free, aims to democratize access to advanced AI technology. While this initiative is commendable, it raises questions about the potential risks of widespread AI access.
Also Read: Volkswagen Integrates ChatGPT into New Models, Redefining the Driving Experience.
