from the OpenAPI initiative is quite stunning . There is plenty of debate over what exactly is its true algorithmic and application value , but the fact that it ’ s been taught 175 billion micro facts from text across the global internet means that we have the first ever single view of all relationships between words , language , point facts , topics and concepts .
“ Just looking at some of the customisation examples that have come from this show how AI can now evolve in the mainstream ; as you pose questions to GPT3 it starts answering and explaining facts , adding new examples and drawing conclusions you may find hard to believe but you know they are informed from 175 billion inputs .”
Q . IN TERMS OF AI , WHAT KEEPS YOU AWAKE AT NIGHT ?
» “ I think we always wonder if AI is heading down the path of the more prosaic Hollywood movies where the computer takes control and one day responds to a request for help with “ Sorry Dave , I cannot do that ”, leaving you wondering how much power has been handed over to silicon . Thankfully computers have power switches and reset buttons , so we need to keep those things and remain in control !
“ Humour aside , the real areas of AI to keep one awake , albeit briefly , are to do with integrity and certainty . As we use AI to support what we do in more parts of our business it ’ s vital to know what error rate and uncertainty lies beneath the veneer of intelligence . Having put high volume products to market in the past , I found
" Your risk and uncertainty can be dominated by the
1 % or even the
0.1 % of edge cases " it never matters what 99 per cent of the product does ; your risk and uncertainty can be dominated by the 1 per cent or even the 0.1 per cent of edge cases . We need to therefore be sure that the benefit model of AI is never outweighed by the hidden cost of any bias , error or uncertainty , which implies we must continually state , measure , validate and report the integrity of AI services .”
Q . WHERE DOES AI GO NEXT IN THE PROPERTY INDUSTRY ?
» “ There are vast collections of static data that span the property industry , many are historical and foretell trends and issues for the future , so I expect some phenomenal new insight and hidden trend models to appear and help us considerably improve the speed of the conveyancing process . “ When you think about shifting society issues and trends in housing development and urban / residential usage patterns , environmental issues – like flood risk and coastal erosion or evolving understanding of land contamination – you can see that future value of property and land are very amenable to AI analysis . Also , dynamically , there is a lot of data moving around the industry telling us how , where , when and why houses sell , mortgages get lent , conveyancing completes , surveys define house valuations and equally interestingly why things go wrong .
“ AI has a lot to offer in explaining what goes well and how to remove the things that don ’ t : I think we all want the process to speed up and there is much to do here .” aimagazine . com 19