Since 2017, Forethought has been constructing chatbots that have become more advanced, smarter and computerized. Now, the company has unveiled their latest progress in progress, enabling general Artificial Intelligence on the platform with a tryout version of SupportGPT.
The product has been made to supply automated customer service answers without human help. Despite the massive publicity created by generative AI, the technology is still in early phases and has its limits. CEO and co-founder Deon Nicholas declared that his firm is aware of this and has engineered SupportGPT’s generative AI to incorporate fewer data than more comprehensive GPT applications. This should supposedly give more precise results.
Nicholas informed TechCrunch that SupportGPT allows their users to start receiving more precise answers to their customer inquiries. Running on OpenAI technology, this has been modified and supplemented by Forethought’s distinctive engineering take on the idea.
He acknowledges that a major problem in the present stage of generative AI is that it sometimes provides the wrong responses. He believes, though, that if we restrict the potential answers the algorithm can provide, we can limit the ‘delusions’ that manifest where AI gives the wrong replies with total certainty.
He explained that hallucinations, or when the AI goes astray, is the major issue in the use of generative AI, which is why they crafted a selection of sophisticated algorithms while utilizing the existing systems that have been created. He then described that this info is given to the generative model either as a suggestion or a guide, and it consequently ends up being closely aligned to the customer’s business procedure.
This feature makes AI more likely to provide answers that make sense. Upwork performed a demonstration showing what happens when a question outside the scope of knowledge is asked–in this case, the weather–the AI recognizes it and brings the conversation back to something it understands. To make sure the AI can provide a suitable response, it is programmed to acknowledge that there is a restricted number of things it can answer. In the given example, when queried regarding the weather, the bot replied that it was not a meteorologist – a reasonable answer to the inquiry – and went on to give samples of questions it can handle.
A major point is providing tailored solutions for each enterprise and organization, so Forethought has introduced a trial version of SupportGPT Playground, where businesses can test SupportGPT by using their own data.
In 2018, Forethought bested the competitors in the TechCrunch Disrupt Startup Battlefield, bringing in an impressive $92 million, as reported on Crunchbase. Most recently, the company closed a Series C funding round at the end of 2021, raising an additional $65 million.
Many companies have begun to benefit from generative AI for commercial purposes. This week, Salesforce announced a trial program for Einstein GPT that provides generative AI abilities thorough their platform. We anticipate numerous other declarations in the near future.
Customer service has relied on chatbots for a while, and this could be a huge step forward in accurate and less irritating conversations with robotic bots. This would help them carry out more precise and useful responses frequently.
Nicholas stated that marrying the LLM (large language model) layer is essential for being able to comprehend a business’ policies and that, when done successfully and on a large scale, it will bring about a significant shift in customer service.