The rise of generative AI is relatively recent and companies still need time to understand the technology, but leading figures in the field are already thinking about the future. Recently, Mustafa Suleyman, co-founder of DeepMind and Inflection AI, said that the next phase after generative AI is interactive AI, an era where machines perform complex tasks on their own by interacting with other AIs and even people. Suleyman said this would allow people to not only obtain information, but also order tasks and services to be carried out on their behalf.
Generative AI is a type of AI technology that can produce various types of content, including text, images, audio, and synthetic data. Although the technology is not completely new, it has generated a lot of excitement in recent months due to the simplicity of new user interfaces that allow high-quality text, graphics and videos to be created in seconds. All of this requires very little effort and allows the user to save costs in some cases. These capabilities have opened perspectives, sometimes radically new, for various areas of human activity.
However, for Mustafa Suleyman, co-founder and CEO of Inflection AI, this is just the beginning. (Previously, he co-founded and led DeepMind, an artificial intelligence company that was acquired by Google for $500 million in 2014.) In a recent interview with MIT Technology Review, Suleyman said the next generation of AI tools will be a step forward. change in the history of our species. According to Suleyman, this new era will be that of what he calls “interactive AI”, in which the possibilities offered by technology will be multiplied by ten, AI tools will be able to easily communicate with each other and perform tasks together.
“The first wave of AI focused on classification. Deep learning showed that we could train a program to classify different types of input data: images, video, audio, language. We are now in the wave of generative AI systems. This phase involves taking this input data and producing new ones. The third wave will be the interactive phase. That’s why I’ve long been betting that conversation is the interface of the future. You know, instead of clicking buttons and typing, you’ll talk to your AI,” Suleyman said. According to him, using AI will be even easier.
Suleyman says that with interactive AI, users will be able to ask AI tools to perform tasks for them and they will carry them out by interacting with other people or other AIs. “This is a big change in what technology can do. This is a very, very profound moment in the history of technology that many people underestimate. Today technology is static. It does, broadly speaking, what you tell it to do. But now technology is going to liven up. In the future, technology will have the potential freedom, if given, to act,” Suleyman said.
Basically, Suleyman says, instead of clicking buttons or typing, users will talk to their AI tools and ask them to take action. “You’ll simply give her a general high-level goal and she’ll use every tool she has to act on it. She will talk to other people, other AIs. It is a real change in the history of our species to create tools with this type of agency,” he explained. When asked about the risks of autonomous AI, Suleyman said it was important to set limits on the technology and ensure it was aligned with human interests.
Many critics feel that generative AI is a misnomer, as these large language model (LLM)-driven programs do not demonstrate intelligence in the same way that humans or animals do. But Suleyman suggests that interactive AI will be closer to artificial intelligence than is often seen in science fiction media. Suleyman’s statements, however, are the subject of criticism, as some believe that the next wave he talks about, that of interactive AI, is already a reality. In fact, OpenAI’s GPT-4 AI model can already perform tasks using other software.
GPT-4 offers advantages over traditional chatbot solutions. It has been provided with the ability to access the Internet. This update allows the template to use over 70 third-party browser plugins, giving users a wider range of features. According to OpenAI, Internet access allows GPT-4 to produce more accurate and current responses, even on topics related to recent events. By taking advantage of the wealth of information available online, the model can provide well-informed and insightful answers that incorporate the most current facts and statistics.
The integration of third-party browser plugins, including widely used platforms like Slack, is expected to improve productivity with AI-based assistance. GPT-4’s capabilities are not that different from what Suleyman describes. However, OpenAI presents GPT-4 as a generative AI. On the web, others accuse Suleyman of simply seeking visibility. “Mustafa Suleyman is more of an activist, more of a marketing specialist (I’ve never seen a real professional so hungry for attention, book tours and media blitzes) and regulation,” the comments read.
“That guy again?” I know Mustafa Suleyman just published a new book, but his PR must be working overtime with all these nonsense statements in various articles. He left Google to start his own startup and is now trying to get the government to restrict independent research in AI. There is nothing altruistic about it,” another reviewer wrote. Suleyman once predicted that everyone will have artificial intelligence assistants in the next five years. His startup Inflection AI launched his own chatbot Pi AI in May that focuses on personal conversations and advice.
According to analysts, we are currently seeing the rise of generative AI tools that go beyond the chat interface popularized by ChatGPT last November. They believe that the next wave of AI startups would allow developers to create applications using AI models and integrate them with external data sources. OpenAI also launched a code interpretation feature for its chatbot in July, prompting Wharton professor Ethan Mollick to say it was “the strongest case yet for a future in which AI is a valuable companion for work.” of sophisticated knowledge”.
Suleyman’s comments come amid concerns that the rise of generative AI is being overestimated. In fact, while generative AI remains a multibillion-dollar industry that is transforming businesses, the buzz around this technology has died down a lot in recent times. Traffic to ChatGPT’s site has declined for the third month in a row, while some ChatGPT rivals, including Jasper and Synthesia, have seen no or declining user growth.
Source: Mustafa Suleyman, Co-Founder and CEO of Inflection AI
And you ?
What is your opinion on the subject?
What do you think of Mustafa Suleyman’s statements?
What do you think of the interactive AI that Mustafa Suleyman talks about?
Is interactive AI really the next step or does it already exist?
What comparison do you make between GPT-4 and the interactive AI mentioned by Mustafa Suleyman?
Is achieving a form of human-level AI faster than expected? There would be a one in two chance that an “AGI” will be designed, tested and publicly announced by 2040.
GPT-4 went from 98% to 2% correct on a simple math problem in a matter of months, study finds ChatGPT and GPT-4 performance is declining
UK judge finds ChatGPT ‘very useful’ after using it to draft decision, says technology has ‘real potential’