Amazon Alexa Inventor Warns Companies Misusing AI: Launches Unlikely AI to Address Industry Challenges
In the fast-evolving realm of artificial intelligence, William Tunstall-Pedoe, renowned for his role in the creation of Amazon's voice assistant, Alexa, has voiced concerns over the current trajectory of AI integration in businesses. Speaking to City AM, Tunstall-Pedoe criticised the widespread yet flawed efforts to embed AI technologies across various sectors, highlighting that many corporate AI initiatives are faltering due to unreliable machine learning (ML) processes.
Having sold his voice assistant startup, Evi Technologies, to Amazon in 2012, Tunstall-Pedoe significantly contributed to the development of Alexa. Now, he embarks on a new journey with the launch of Unlikely AI, a deep tech startup aimed at overcoming prevalent AI challenges. "A lot of these projects are failing because the ML piece doesn’t work well enough. It isn’t trustworthy, goes awry in ways that could lead to cost implications or fall outside regulatory frameworks, or worse, harm the brand," Tunstall-Pedoe stated.
Based in London, Unlikely AI is positioning itself as a beacon for solving critical issues such as bias, hallucination, accuracy, and trustworthiness in large language models (LLMs). The startup’s ambitions are particularly focused on high-risk industries like finance and healthcare, where he aims to build an "AI platform that’s trustworthy and explainable, enabling businesses to apply AI confidently."
Backing this venture is a robust $20 million seed investment, led by prominent investors Amadeus Capital and Octopus Ventures. The financial support underscores the confidence in Unlikely AI’s mission to reshape how businesses interact with AI technologies.
As Unlikely AI quietly nurtures its innovations, it is simultaneously enhancing its organisational structure with strategic hires. The company recently appointed Fred Becker, previously part of the scaling team at Skype, as its chief administrative officer, and Tom Mason, formerly of Stability AI, has joined as the chief technology officer. Mason remarked that Unlikely AI’s technology is engineered to tackle sophisticated challenges, likening it to automating claims processing in a predictably accurate manner, setting itself apart from conventional language models.
With its plans under wraps for the most part, Unlikely AI is poised to unveil its commercial operations by the second quarter of next year. As industries await further disclosures, the company’s forward-thinking approach aims to deliver a transformative solution for AI integration challenges that have perplexed businesses worldwide.
Through its pioneering developments, Unlikely AI is committed to meeting the needs of industries that require precision and reliability, with the potential to redefine the landscape of AI applications. As sectors continue to experiment with machine learning technologies, Tunstall-Pedoe's insights and Unlikely AI's advancements may well inform future strategic directions in AI deployment.
Source: Noah Wire Services