Trending Articles
Trending Topics
Trending
The AI Alliance celebrates its first birthday today, with 5 December 2023 marking the date that IBM, Meta and over 50 other tech companies combined forces to create a more open environment. So what has it achieved in that year? And what happens next?
At an IBM Research event that took place on the anniversary, we asked Anthony Annunziata, IBM AI Open Strategy Director and AI Alliance Lead, for his thoughts. “Things have gone really well,” he told us.
“We have brought together more than 140 organisations from all over the world, with people from all sorts of technical and non-technical backgrounds working together to help make AI more trusted, safer and more useful. So it’s very inspiring.”
Agata Ferretti, IBM’s AI Alliance Program Lead for Europe, added that “in Europe, we have a growing community of people that are interested in building technology that is safe and is open and it’s trusted for the good – not of the few, but of the many.”
First, let’s summarise why the AI Alliance was set up in the first place. Several storms raged in the AI universe in late 2023, whether it was Nvidia’s growing dominance or a knee-jerk and confused reaction from governments wondering how to deal with this growing force.
IBM and its AI Alliance partners realised that they would be more powerful together than battling alone. Otherwise, Nvidia and OpenAI could be the ones who set the standards, while governments around the world listened to hundreds of different voices and – perhaps – regulated the nascent industry out of existence.
Or, worse, were set down a path set by Nvidia and other AI gorillas.
There was another problem too. The big names, including Nvidia, Microsoft and OpenAI, relied on proprietary technologies rather than open versions. There was, and remains, a “trust us with your data, we’ll look after it we promise” approach.
Not that the AI Alliance puts it in such negative terms. It talks of “a global collaboration committed to building, enabling and advocating for open innovation in AI across the technology and societal landscape”. But that amounts to much the same thing, I would argue.
The AI Alliance adds one further clause: “… to ensure that AI technology is accessible, trusted and beneficial to everyone”. And who can argue with that?
Let’s summarise the AI Alliance’s achievements with a quick list:
That’s not a bad set of achievements for a one-year-old. But perhaps more importantly, many companies and organisations are now on board that might otherwise have gone all-in with Nvidia.
If you’ll forgive us another list, this group includes:
“Today we are committing to scale up the breadth of our reach and impact by 10x in 2025,” read a statement prepared by IBM. It announced two initiatives that mark the direction of travel.
First, there’s the Trust and Safety Evaluation Initiative (TSEI). This will include an assessment of taxonomy and domain-specific risks, the creation of benchmarks for taxonomies and an AI trust and safety reference stack.
It’s also announced the Open Trusted Data Initiative (OTDI), which will provide guidance on how to build and deploy AI systems so that they can be trusted.
Looking at the map of existing AI Alliance members, there is also a need to expand its reach beyond North America. For example, currently there are only two members of the alliance in Africa.
“It’s not perfect,” said Annunziata. But he added that “we’re about to have one of the major universities join and we have a couple of projects from one from another startup in African languages and modelling that are going to be kicking into gear in the New Year.”
In terms of sheer numbers, Europe also has some catching up to do. But Ferretti is optimistic. “There is a lot of interest, especially research, but also startups are really keen to [leverage] the power of the open because often there is a lack of resources in certain environments.”
But Ferretti was keen to point that it was a two-way street. “It’s really an opportunity for them, but also they are ready and willing to contribute to share their knowledge, which I think is very impressive.”
So what will happen next for the AI Alliance? Certainly more members, more initiatives, more engagement with governments. Because when we asked Annunziata which one organisation he would want to join the AI Alliance, that’s the area he chose.
