OpenAI just announced its new GPT-4o model: 5 key takeaways for education institutions

On 13 May, OpenAI announced the release of its new GPT-4o model, marking a significant leap in the field of artificial intelligence. This latest iteration of the GPT series introduces a host of improvements including a new voice and video mode that OpenAI’s CEO Sam Altman described as the “best computer interface I’ve ever used”. The new model is available to both the free and paid versions of ChatGTP, meaning all users will have access to OpenAI’s most advanced technology. The improvements are aimed at enhancing performance, usability and utility across various applications with major implications for the education industry. 

Here are 5 key takeaways for education institutions:

‍

‍1. AI is becoming “Omni-Modal”, increasing usability and inclusivity

Before GPT-4o  (where “o” stands for “omni”), using Voice Mode required  a pipeline of three separate models: one to transcribe audio to text, GPT-3.5 or GPT-4 to process the text, and another to convert the text back to audio. This setup meant that GPT-4 could not directly understand tone, distinguish between multiple speakers, or recognise background noises. It also could not produce sounds like laughter or emotional expressions, leading to a loss of important contextual information. GPT-4o now directly supports diverse input and output formats - such as text, speech, video and images. 

We expect this update to have a significant impact on the adoption of AI in education. The multi-modal capabilities can ensure that students with disabilities, such as visual or hearing impairments, can access and engage with AI and educational content more effectively. Additionally, the feature boosts the possibility to create more dynamic and interactive learning experiences, from virtual labs to real-time simulations. Looking forward, omni-modality has the potential to result in a fundamental paradigm shift that could change the way humans, and therefore students, relate to computers and each other.

‍

2. AI is becoming more accessible for non-English speakers 

GPT-4o significantly improves speech recognition and translation performance over previous models across over 50 languages, particularly for lower-resourced languages.

‍

‍

As education institutions look to integrate more live transcription and translation capabilities, GTP-4o presents a number of opportunities to  bridge language barriers in international classrooms or to faster and better localise education offering to new markets.

‍

3. AI is getting closer to being on everyone’s desk

GTP-4o’s native vision and audio capabilities also present significant integration improvements. Practically, this is the ability for the AI to use a phone’s camera or computer screen. In a live demo, the OpenAI team showed ChatGPT an equation they had just written on a piece of paper and asked  for help solving it. Instead of providing the answer directly, the AI offered guidance and walked them through the problem step-by-step. The model  was also able to track and respond to the changes made in real-time, as a result of a new continuous live chat feature.

OpenAI has also launched a ChatGPT desktop app, able to view the user’s screen and act as a real-time personal assistant and coach. This development, along with the recently announced customisable chatbot builder and enterprise ChatGPT solution, marks a significant step toward making AI tools widely accessible as fundamental components of both workplace and university infrastructures.

‍

4. The need for ethics and governance standards becomes more pressing

GPT-4o’s new multi-modal and always-on feature sets  present a variety of novel risks that come in addition to the widely-discussed concerns around ethics and governance of AI in universities and corporates. 

As AI technologies  advance, the need for establishing clear policies and governance frameworks becomes ever more crucial  to ensure that the use of new models aligns with ethical standards and protects all stakeholders. Organisations have a responsibility to promote transparency, fairness, and accountability in AI applications, fostering trust and ensuring that the benefits of advanced AI technologies are realised responsibly and equitably across communities.

‍

5. Innovation cycles are shortening, calling for new approaches to AI adoption and governance 

The increased speed of technological advancement is reducing the temporal relevance of numerous processes, products, and best practices. As a result, education institutions are at risk of constantly lagging behind innovation. 

To solve this dilemma, organisations should look at building on the capacity to shift to proactive approaches to both AI adoption and  governance. These approaches should be focused on the outcomes and processes enabled by technology, instead of the technological innovations themselves.

‍