The Global AI Newsletter
Tuesday, March 31 2020
In partnership with Henkel X Ventures
Welcome to the Global AI Newsletter – the first data-driven digest of how the race to develop and deploy artificial intelligence is unfolding globally.
This month: as the coronavirus crisis unfolds, we’ve delved into two of our index’s live data sources to get a sense of what impact the pandemic is already having on AI.
Plus: how a lack of trust may stop an effective AI response to c-19, and more details on how you can take part in the Tortoise AI Summit in May.
Have questions about our data? Or just want to get in touch? Email me at email@example.com.
AI after C19: Slowing down?
What we’ve found: AI startups are getting caught up in the looming global recession. The number of funding rounds announced by companies developing AI tech has now dipped to new lows, according to data from Crunchbase, an investment platform we use for our index’s commercial indicators.
The bright side: You’d be hard pressed to find any business that’s not suffering at the moment (aside from Zoom) – but it looks like AI companies are actually faring relatively less badly. The share of cash announced from funding rounds going to AI ventures has been on the up in recent months, reaching an all time high in March.
One caveat: March’s data is slightly skewed by the announcement of Waymo, Google’s self-driving car unit, raising a massive $2.25bn
And one other thing: Danny Crichton at Techcrunch – the media outlet that owns Crunchbase – has suggested companies are holding off on any funding announcements for fear of having their good news overshadowed by all the covid-19 headlines.
Bottom line: With such powerful tech, it’s easy to think the AI industry is invincible – but the coronavirus crisis might be about to painfully reveal just how untrue that is.
AI after C19: Coding the cure?
What we’ve found: Data scientists, coders and academics from the countries hit most badly by the covid-19 pandemic are flooding the web with data and code for analysing the virus. Universities and research institutions from Italy, which is otherwise one of the lowest ranked European countries in our index, are behind the second highest number of apps hosted on GitHub that track and disperse data on coronavirus.
How we got the data: By using the GitHub API to find repositories – collections of code and data – that were tagged with coronavirus-related terms like ‘covid19’, ‘wuhan-pneumonia’ or ‘SARS-CoV-2’. That gave us 1674 in total, most of which were created in March. For our country counts, we’ve only looked at those repos owned by an official organisation in that country, not just those repos of any user.
What about AI? We checked if any of the covid-19 repos also included tags that indicated the use of artificial intelligence explicitly. Less than 3% reported a combination of these tags, which probably indicates that most of the coding activity is concentrated on tracking the virus’s spread and its effects rather than predicting the future.
Special mention: John Hopkins University in the US, whose GitHub repo for daily covid-19 data is by far and away the most popular. It’s got over 18,000 ‘stars’ – the GitHub equivalent of a ‘like’ on social media – compared to the average among our covid-19 repos of 187.
Bottom line: When crisis strikes, data scientists and coders aren’t far behind – but the full extent of how the open source community is actually modelling all this data remains to be seen.
A message from Henkel X Ventures
“We’re excited to be a partner in The Global AI Index – and we look forward to building a community of like-minded businesses and policy makers to accelerate change in the field of AI.”
— Marius Swart, Co-founder of Henkel X Ventures.
AI after C19: In data we trust?
What we’re thinking: Covid-19 is profoundly exposing our dependence on data – but with trust in AI varying dramatically country-to-country, some governments may have a harder time leveraging such statistics for public health.
What we know already: We’ve gathered data on public perceptions of AI for our index, offering clues to the kinds of countries who might suffer from such ‘trust disadvantages’. The older and richer a country’s population, the less favourable that country’s attitudes are to AI, we’ve found.
An AI model a day…: Right now South Korea offers a tricky case in point. On the one hand, the country has rolled out by far and away the biggest mass programme of testing in the wake of covid-19. On the other, the country had the second lowest number of respondents to a global study by IPSOS in 2018 reporting they trusted artificial intelligence at just 17%.
What we’re reading
On the frontline: An overview of which AI companies are deploying their tech to fight covid-19 / Read more
No silver bullet: AI may help us address coronavirus – but equally coronavirus will expose AI’s shortfalls / Read more
Modelling 101: Why your covid-19 exponential fits are probably wrong / Read more
Mind reading: Researchers in the US have developed AI that can turn brain activity into text / Read more
Ahoy: The startup using AI to cut down on crime at sea by tracking suspicious vessels / Read more
Hiring freeze: There may be such a thing as too many data scientists / Read more
Research: Alex Clark, Roberta Sgariglia
Editing: Alexandra Mousavizadeh, Alexi Mostrous
Save the date: Tortoise Summit AI
What we’re planning: A virtual summit bringing together industry leaders, tech founders, policy makers and Tortoise members to discuss the role of AI in a post-covid-19 world. It coincides with a refresh of our index data to help inform how coronavirus has shifted AI innovation, spending and research across public and private sectors.
The strands: We’ll be asking: after c-19…
- Will companies embrace automation and move towards less labour-intensive workplaces?
- Will AI tech increasingly prioritise health security over privacy and civil liberties?
- Which governments have pivoted spending fastest towards health-based AI since the crisis emerged?
When: Friday, May 15th / 8am to 2pm GMT.
And who: Sessions and breakouts will be chaired by James Harding, co-founder of Tortoise and former head of BBC News, Alexandra Mousavizadeh, director of Tortoise Intelligence, as well as Tortoise editors including Alexi Mostrous, Matt d’Ancona, Giles Whittell and Merope Mills.
How to take part: More details soon