EVERYTHING that RISES…
Angelo Dalli is an AI veteran who believes its time has come
He speaks to BLOCK magazine about the recent surge in AI, and how it could converge with other emerging tech, like blockchain.
Intelligence means different things to different people. No matter how you define it, there’s no denying that Artificial Intelligence has somehow gone from the stuff of science fiction to a revolution that’s not only possible, but is looking increasingly likely. And there’s no better man to walk us through the story so far, and to give us a glimpse into what’s next, than AI luminary Angelo Dalli.
His jovial demeanour and the sheer breadth of his interests belie a true commitment to the cause of Artificial Intelligence done right, and a level of associated expertise that very few can match. Angelo Dalli has been sinking his teeth into innovative tech since 1995. His work in the fields of search engine technology, intelligent transport, big data, online gaming – and even language – has left him with an enviable track record. He’s also made his mark as an entrepreneur, with several winning exits already under his belt.
Before picking Angelo’s brain on the intricacies of Artificial Intelligence I ask him how he defines intelligence, period. He explains that the picture of AI created by popular culture isn’t always helpful. “You hear a lot of misleading claims that we’re creating an artificial brain,” he tells me. “There is this system called neural nets in AI. They are a crude approximation of how the brain works. There is a little inspiration and grounding from biology, but it’s very different from how the brain works. Take a one centimetre cube of brain matter and try to join it to another similar size cube by tracing all the connections between them – there isn’t enough computing capacity in the world yet to do it. I think we are very far away, although I do believe that at some point computers will be on par with human intelligence”.
That particular level of AI is more specifically referred to as Artificial General Intelligence, or AGI – and it is still a way down the road. “There are few things in life that are built overnight,” says Angelo. “People work on individual problems, and bit-by-bit a lot of solutions are created which will eventually be ‘chained’ together”.
I ask him how he decided to invest his energy in AI in the first place. “It was
circumstantial that I got into AI,” Angelo explains. “The University of Malta offered an AI degree – a rarity in itself for that time. With the hardware available at the time we could still do facial recognition, even in 1998. The problem back then was that there wasn’t enough data. While I learnt all the fundamentals, the application areas of AI were still very limited. There weren’t many smart phones, you couldn’t download much and there weren’t many digital photos. They were a little ahead of their time”.
“There was quite a negative attitude towards AI though,” Angelo elaborates. “No one really wanted to work in it, in fact. A lot of this stuff you hear about deep learning started off in Switzerland and Canada, there was a small group of people who kept it going, and it survived. It had a bad reputation because it was over-hyped. A lot was promised that wasn’t delivered. The difference now is that it can deliver.”
I ask him which factors have changed over time, enabling AI to go from tall order territory to the realm of accelerating likelihood. Angelo highlights two primary factors that held back AI in the 80s and 90s, but have since tipped in its favour. These are processing power, and the availability of data. The usefulness of processing power is relatively clear, but less obvious is the impact of data. Angelo explains that intelligent behaviour doesn’t manifest artificially unless and until a critical threshold is reached, in terms of the availability of a suitable ‘learning environment’ made up of critical data.
“In order to ‘train’ AI you need to have a lot of data, but no one knew the minimum limit beyond which it becomes superhuman,” Angelo explains. “The problem was that the guys in 80s and 90s got it right but didn’t have enough data. The most cutting-edge systems are using algorithms invented in the 80s, and are more or less unchanged. What has changed is the amount of data. It took us twenty or thirty years to realise this. Once you have a certain amount of data, all of a sudden you get interesting behaviour – this is what has driven the new wave”. For instance, the amount of images that populate the web is far greater in 2018 than it was in 1998. That massive increase truly is useful if you’re trying to get a machine to recognize human faces and tell them apart. While it’s hard to say exactly where the ‘magical data threshold’ is, it’s there.
In essence, this is why Angelo feels like his time has finally come. He’s spent a large part of his life waiting for technology to catch up to his ambitions, and now it’s a matter of seizing upon a hard-won opportunity to be part of the next watershed moment in human history. To Angelo, the emergence of AI will be equally as seminal as the impact of the steam engine or the plough.
He emphasizes the importance of an ethical approach towards artificial intelligence, however. The immense power that stands to be harnessed ought to be deployed in the right way, freeing human beings up so they may achieve their true potential. However, Angelo concedes that AI could cause a great deal of harm if used carelessly or maliciously.
“I am completely convinced that AI is going to change human society – the next industrial revolution will be powered by it,” claims Angelo. “This is why I believe research needs to be done in an ethical way that impacts society positively. I believe AI in its current form should focus more on assisting people rather than replacing them. I am very much in favour of light regulation. What I would be uncomfortable with is any weaponised use of AI, without human control. I think it is vital that an ethical framework should guide AI research, as one bad apple could spoil the whole thing”.
I ask Angelo if AI plays well with other emerging technologies, like IoT and blockchain. Based on what’s already been established about the usefulness of data, there’s an obvious overlap between AI and IoT. With blockchain the overlap is less obvious, and Angelo seems to think the two technologies can develop in parallel.
“AI will use blockchain mostly as a useful tool, whenever it needs to record something immutably,” Angelo tells me. “If you have really sophisticated systems, how will you pay the AI if it starts charging you? I think this will become more common, and it will enable transactions to happen more efficiently. I think as the amount of transactions grows you will need AI algorithms to help you sort out the mass of data – to help people make sense of it”.
As AI continues on its ascendancy, Angelo Dalli is poised to be in the thick of things. While it is not the only exciting emerging technology out there, it will certainly have a say in how we’ll live our lives in the nearer-than-you-think future. Also, just as AI had to bide its time and wait for processing and data innovations to occur, it now stands to enhance other technologies and in turn enable further technological leaps – for everything that rises must converge.
Come and meet Angelo Dalli at Malta AI & Blockchain Summit (22nd – 24th May), where he’ll be speaking at our conferences on all things related to AI!