While OpenAI’s ChatGPT and similar artificial intelligence (AI) chatbots continue to garner attention and intrigue, artificial intelligence is already taking the next big leaps.
Like India-based AI startup Writesonic, which launched its AI chatbot Chatsonic just a week after the breakthrough launch of ChatGPT last year in November, terming it a ‘ChatGPT Alternative Built With Superpowers’, has plans to offer bots customised for an organisation’s individual needs. “Our new offering, Botsonic, a no-code AI chatbot platform, will enable enterprises and companies to create custom chatbots similar to ChatGPT, tailored specifically to their unique data and requirements,” Writesonic’s Delhi-based founder Sumanyou Garg told FE in an interview.
Another feature it plans to introduce is the document-aware content generation, which allows Chatsonic to generate content that intelligently utilises an organisation’s existing knowledge base. By integrating with an organisation’s internal documents and resources, Chatsonic will be able to create contextually relevant and informed content that aligns with the organisation’s specific expertise, industry, and goals, Garg said.
Greater personalisation is also on the cards as the company “aims to offer hyper-personalised content creation across various formats, including blogs, presentations, emails, reports, and videos,” the Writesonic founder said, adding, “As AI continues to evolve, we are also constantly exploring new features to keep pace with technological advancements and meet the ever-changing needs of our users.”
Chatsonic claims it has addressed limitations of the OpenAI’s widely popular chatbot, which Microsoft has integrated with its Bing search engine. These include “information limitation as its knowledge base only extended up to September 2021, the lack of speech-to-text and text-to-speech capabilities, and lack of multimodality,” the Chatsonic maker said. Last month, Opera integrated ChatGPT and Chatsonic into its desktop browsers—Opera and Opera GX.
Content creation is the foremost utility of Chatsonic, said Garg. However, wouldn’t it pose a threat to content creators and has the potential to make their job obsolete? On this, Garg bats for the ‘productivity’ quotient, and says, “It functions more as a support for humans and improves their productivity, rather than taking them down.”
What about inaccurate responses or controversial content like hate speech? To address that, “we have our toxicity detection system. So if you ask it to say something about Hitler, it would straightaway decline as it is against our system. It is the same for images,” said Garg.
However, AI is expanding at an exponential rate, and so is the expanse of content that can be termed controversial and dangerous. On how tech keeps pace, Garg said there are multiple aspects to it. “We keep retraining our toxicity detection model and it is updated every few weeks. Secondly, the point where AI is currently at, 100% accuracy is not guaranteed. Although there is a decent level of guarantee, with models improving, we can expect it to get there where it is smart enough to automatically deny requests without the need of training it,” he said.
Also read: WhatsApp chat transfer feature starts rolling out, here’s what it is all about
However, the techpreneur does feel that at the rate at which this technology is evolving, some kind of regulation is needed. Notably, multiple countries are deliberating upon some forms of regulation for AI. “But these should not be so strict, as a lot of good things will also come out of it,” Garg opined.
We are already at a time when generative AI, the technology behind AI chatbots, has set the tech and non-tech worlds abuzz. Speaking on what could be the next big thing in the field, Garg feels “autonomous GPT or similar self-improving generative models can be one of the next big developments in AI.” It is an open-source application, which displays the abilities of the GPT-4 language model. Notably, autonomous GPT can function autonomously, hence limiting the need for human-machine interactions. “Auto GPT would represent a significant advancement in the capabilities of AI language models like GPT-3. A preliminary version of this is already available on GitHub as an experiment,” Garg said.
Source link