In the latest episode of Unconfuse Me with Bill Gates, OpenAI chief Sam Altman pointed out that OpenAI is on ‘this long, continuous curve’ to create newer and better models. He highlighted the importance of multimodality as the key aspect of GPT-5 that enables it to process video input and generate new videos. 

Altman believes that soon, AI will be able to handle more complex tasks, leading to a boost in productivity. “You can imagine a little agent that says, ‘Go, write this whole program for me’. It may ask you a few questions along the way, but it won’t just be writing a few functions at a time – it’ll enable a bunch of new stuff,” Altman said, suggesting that it will handle even more complex tasks.

“Someday, maybe there’s an AI to which you can say, ‘Start and run this company for me’. And then someday, there may be an AI to which you can say, ‘Go discover new Physics [laws] for me’,” he quipped.

As of now, several foundational models have achieved the capabilities of GPT-4. Google is likely to launch Gemini Ultra at any moment. Meanwhile, Mistral CEO Arthur Mensch announced on French national radio that the company will unveil an open-source GPT-4-level model in 2024. 

“Will Gemini Ultra launch before GPT-5 or vice-versa? My bet is on Gemini Ultra… and I suspect it beats GPT-4 on reasoning,” read Abacus AI chief Bindu Reddy’s post on X.  

Notably, Mistral Medium has gathered 6000+ votes and is showing remarkable performance, reaching the level of Claude. “Mixtral will take over GPT-4 this year. Today, it’s the only open-source model at the top in Chatbot Arena (following GPT-4, Claude, and Mistral Medium) and the smallest one with 7B instructions. It’s even better than Google’s Gemini Pro,” exclaimed AI expert Santiago.

Facts and claims like these keep OpenAI on its toes.

What Next – GPT-5?

In conversation with Gates, Altman spoke at length about GPT-5, emphasising on customisation and personalisation. “The ability to know about you, your email, your calendar, how you like appointments booked, connected to other outside data sources—all of that. Those will be some of the most important areas of improvement,” said Altman. 

Furthermore, he claimed that GPT-5 would have much better reasoning capabilities than GPT-4. “GPT-4 can reason in only extremely limited ways. Also, reliability is a concern. If you ask GPT-4 most questions 10,000 times, one of those 10,000 is probably pretty good, but it doesn’t always know which one. You’d like to get the best response of 10,000 each time,” said Altman. 

Along with new models, OpenAI might soon start focusing on specific verticals to serve its customers. “Coding is probably the only area for which we’re most excited about productivity gain today. It’s deployed massively and used at scale at this point. Healthcare and education are also coming up that curve, and we’re very excited about them too,” said Altman. 

Meanwhile, OpenAI recently introduced the GPT Store. This could potentially aid OpenAI in developing GPT-5, as they will receive labelled data at scale from customers creating GPTs using their personal data. Interestingly, there is no mention in the blog that OpenAI won’t use GPT Store data to train its models.

OpenAI is All About Action 

The OpenAI team is quite active on X, asking for customer feedback to improve its models. Recently, OpenAI president Greg Brockman posed a question on X, “How has ChatGPT changed your life?” A user named Aaron Stormerr shared the profound impact ChatGPT has had on his academic journey as a blind computer science student.

Similarly, OpenAI developer relations head Logan Kilpatrick asked, “Who is building the most useful/coolest products with OpenAI API? I want to spend more time in 2024 hearing from builders about what we can do to support them/you.” 

Clearly, the OpenAI mantra is all about releasing a subpar product or a feature, and improving them on the sideline with continuous consumer feedback, plus shipping them as quickly as possible. So, the company is as good as its next model – obviously, drawing inspiration from Walt Disney who used to say to his team: “We’re only as good as our next picture.”

The post OpenAI is as Good as its Next Model  appeared first on Analytics India Magazine.