Today marks the official product launch of federated AI development platform Sherpa.ai, in a notable advance for automation that learns from multiple data sources without information ever leaving its original location.
Federated AI could massively help out training AI to solve intractable societal challenges, by allowing data collection to happen without undermining privacy rights of those providing.
Fears surrounding privacy undermine tech's credibility with the public, but Basque Country/Silicon Valley-HQed Sherpa.ai has created an algorithm engineering suite that doesn't require any private data to be relayed away from its source.
Data privacy laws in Europe became more widely prevalent with GDPR's introduction, and California is among the US states to have followed suit. More stringent privacy rules affect some the more lucrative AI end points, notably in health care and finance.
Sherpa.ai aims to be the most advanced platform out there in terms of data privacy, bringing federated AI to global multinationals. A second benefit is reductions to carbon emissions. Most AI is trained from huge cloud data servers and these usually have to execute consistently until the algorithm understands its workload.
Sherpa.ai's management includes a true heavyweight of consumer AI.
Director of AI strategy Tom Gruber previously helped to pioneer the voice-activated smart assistant Siri, now ubiquitous in Apple products. Gruber is joined by Sherpa.ai founding CEO Xabi Uribe-Etxebarria, an Oxford business school graduate, along with former US director of science and tech policy Thomas Kalil.
The product launch was marked at a special keynote event attended by Sherpa.ai's data privacy stakeholders. Early agreements to use the federated AI creation technology have been concluded with the US National Institutes of Health, European telecoms group Telefónica and accounting firm KPMG, among others.