"Open sourcing artificial intelligence (AI) can help combat concentrations of capital and power that currently define its development, while nascent assurance practices need regulation to define what "good" looks like. Speaking at trade association TechUK's ninth annual Digital Ethics Summit in December, panellists discussed various dynamics at play in the development of AI technologies, including the under-utilisation of open source approaches, the need for AI assurance to be continuous and iterative, and the extent to which regulation is needed to inform current assurance practices."
"During the previous two summit events - held in December 2023 and 2024 - delegates stressed the need for well-intentioned ethical AI principles to be translated into concrete practical measures, and highlighted the need for any regulation to recognise the socio-technical nature of AI that has the potential to produce greater inequality and concentrations of power. A major theme of these previous discussions was who dictates and controls how technologies are developed and deployed, and who gets to lead discussions around what is considered "ethical"."
Open sourcing AI can reduce concentrations of capital and power and broaden participation in AI development, while nascent assurance practices require regulation to define what "good" means. Most AI models exist on a spectrum between open and closed rather than fitting a binary, and genuinely open-source models are exceedingly rare. AI assurance needs to be continuous and iterative, with regulation shaping emerging assurance practices. Ethical AI principles require translation into concrete practical measures, and regulation should recognise AI's socio-technical nature and its potential to produce greater inequality and concentrated power. Efforts to democratise development include exploring open models and ecosystems, though open-source approaches remain under-utilised.
Read at ComputerWeekly.com
Unable to calculate read time
Collection
[
|
...
]