Two years ago, OpenAI introduced ChatGPT, a subscription-based conversational AI model that can provide AI services at any monthly cost. Although its shortcomings include poor training data quality and persistent hallucinations issues, ChatGPT quickly gained popularity – becoming one of the world’s most-used platforms.
However, the company hasn’t updated their model in quite some time; CEO Sam Altman recently provided more detail in an AMA on Reddit.
1. AI Agents
AI agents offer users new ways to engage with generative models. For instance, if conducting market research for your next product launch is required, an agent could research Ideal Customer Profiles and return a detailed report with data, documents and presentation slides.
Another option would be to hire an agent to manage email correspondence and manage price negotiations with clients. You can set an expected range for pricing negotiations before handing it off to an agent so you have time for other matters.
GPT-5 should also enhance other aspects of ChatGPT, such as reducing “hallucinations”, which occur when AI generates incorrect or nonsensical information. Furthermore, its model will be more explainable and efficient, requiring less computing power than previous versions – making it simpler to use for tasks requiring complex answers or multiple steps.
2. Video Capabilities
ChatGPT has already revolutionized how video content is created and consumed. The GPT-4o model, created to enhance natural language processing – one of its major components – has made hosted video solutions like Zoom simpler for its users to utilize.
GPT-4o with vision, the latest addition to our model, integrates image and voice input. Users can prompt AI with images such as coffee-making setup, such as when ordering coffee from their local cafe. AI then responds like a virtual barista and answers their queries accordingly. This feature encourages people to treat AI more like an individual and make it feel present in their lives.
Currently available to ChatGPT Plus and Pro subscribers, Enterprise and Edu customers will gain access in January. CEO Sam Altman demonstrated this feature on 60 Minutes where an AI assessed drawings of anatomy while being asked questions by journalist. Though still developing (it made an error on geometry problem), it was impressive seeing an AI recognize images drawn by humans in real time.
3. Personalized Data Sources
Future projections call for ChatGPT to process more data from various sources, for instance learning more about its users so it can provide personalized responses more effectively – this feature could prove especially valuable in fields requiring specialized knowledge such as healthcare, finance or law.
Predictive personalization, which allows ChatGPT to anticipate customer needs and wants before they’re explicitly expressed, should help increase engagement and conversion rates for marketers.
ChatGPT can respond to prompts with short persuasive messages tailored to an individual’s psychological profile (e.g., writing persuasive ads for Nike sneakers for people who have an emphasis on prevention” or writing political ads advocating climate action for people scoring highly on fairness), already revolutionizing workplaces; one blind study found licensed health care professionals who reported finding chatbot responses more accurate and empathic than physician answers when engaging patients directly.
4. Artificial General Intelligence
As AI researchers work to advance large language models, their scope of functionality will expand accordingly. This could include more explainable AI, reduced biases in training data and more efficient algorithms that require less computing power.
Experts predict this trend will lead to text-to-action AI that allows users to instruct their computer to complete tasks for them. Unlike AI systems which can already manage many processes on your PC, these bots would perform specific tasks that you ask of it.
Before this can happen, however, considerable work needs to be completed. Most importantly, we must establish what constitutes Artificial General Intelligence (AGI). Unfortunately, many aspects often associated with AGI are neither economically beneficial or can only be realized indirectly–like artistic creativity and emotional intelligence.