OpenAI and Meta are at the tip of announcing next-gen innovations in the field of artificial intelligence. The upcoming models of machine intelligence are expected to utterly transform cognitive abilities of these machines by granting them exceptional logical reasoning and tactical planning. The top of the executives from both the companies recently have given hints about the emerging soon upgraded versions of their flagship large language models.
These advanced systems, which play the key role in various kinds of generative AI applications such as ChatGPT, enable machines to move ahead to the border of human intelligence and simultaneously take a step close to the highly coveted artificial super intelligence.
Meta vowed to offer Llama 3 in the forthcoming weeks, which is also the time period when OpenAI, with support from Microsoft, is expected to unveil the next-gen model, called GPT-5. Its announcement about this new model is close and may come soon.
As indicated by OpenAI’s chief operating officer Brad Lightcap in a statement shared with the Financial Times, the upcoming release of GPT will be showcasing more significant improvements in being able to address difficult tasks especially when it comes to reason.
Meta and Open AI’s advancements are only part of a more significant industry-wide trend in tech, where companies like Google, Anthropic and Cohere are all working on smaller detailed language models to be released this year. This wave reportedly evidences the never-ending race among the tech leaders to develop more advanced generative AI modalities that have the potential to mimic the quality of individuals’ produced content such as (text, images, code, or video). The race is getting tighter as innovation hardens and transforms in this industry.
The capabilities of logical reasoning and planning are the most advanced ones in realizing what is AI researchers called as “artificial general intelligence” (AGI). It is the human-level intelligence. These advancements are basic in nature because they give the capability to virtual assistants and chatbots to do connected tasks and predict the outcomes of what they have done.
Due to the fact that their struggle is hard to tell an inquiry apart from their retention of information over a longer period, they are known for “stupid” mistakes, according to him. In AI model, inclusion of reasoning involves a process where the AI examines preceding probable reasons and then, by applying a sequence of actions, forms a cognitive conception of the foreseen outcomes of it.
He highlighted this component as a huge gap in machine learning capabilities that researchers are working on to graduate AI from current level to the next level. LeCun touched upon current projects that are aimed at creating AI “agents” capable of tasks, such as providing a means of transport from an individual’s office in Paris to another in New York, and taking care of travel related aspects at the airport.
Meta is going to fit its new AI model in WhatsApp and their Ray-Ban wearable devices. The manufacturer is basically prepared to introduce Llama 3 available in different sizes and optimized for diverse applications and devices during the next couple of months.
