Artificial intelligence sees more funding, but needs more people and better data
The state of artificial intelligence is promising, and it’s increasingly ready for real-life businesses. But there are talent shortages, a lack of diversity in the field, and concerns about data management that are fueling increasingly sophisticated algorithms.
These are some of the observations of Nathan Benaich and Ian Hogarth, leading investors in artificial intelligence, who released their fourth annual and densely packed edition.State of AI“report examining developments in the field over the past year. Although the report focuses on AI universities and specific advancements in medicine and other fields, important developments are raised for those who seek to leverage AI and machine learning to move forward in building smart businesses. “AI alignment efforts underfunded by key organizations advancing the field global AI, as well as concerns about the datasets used to train AI models and bias in model evaluation benchmarks, raise important questions about how best to track the progress of AI. AI AI systems with rapidly evolving capabilities, âsay Benaich and Hogarth.
Some notable developments in AI over the past year are as follows:
- AI is now part of important real life scenarios, including being applied to critical infrastructure such as national power grids, automated optimization of supermarket warehousing, drug discovery and healthcare.
- “Transformers“, a deep learning architecture based on a neural network, has become a general purpose architecture for machine learning, increasingly applied to natural language processing (NLP) and computer vision.
- Other developments mentioned included the rise of computer vision self-monitoring which requires less training, and “text-less” natural language processing based on computer vision. Generative Spoken Language Modeling (GSLM), which enables the “task of learning vocal representations directly from raw audio without any labels or text”.
- This year, record funding has gone to AI startups and IPOs of data infrastructure and cybersecurity companies that are helping companies retool for the first era of AI.
AI talent is a growing concern, as well as an area of ââopportunity. âComputer researchers, software developers, mathematicians, statisticians and data scientists have seen their jobs grow far beyond that of the general workforce,â Benaich and Hogarth say. âComputer Science and Engineering were the fastest growing undergraduate degrees between 2015 and 2018, accounting for 10.2% of all four-year degrees awarded in 2018. Their number increased by 34% and 25%. % respectively during the period, while the number of other degrees awarded increased by 4.5% on average. “
Globally, Brazil and India are leading the way in AI job growth, hiring more than three times as many AI talent today than in 2017, matching or exceeding the growth of the hiring in Canada and the United States, they add.
Data on gender and racial diversity within American organizations differ dramatically between technical and non-technical teams in Benaich and Hogarth states. There is “a massive lack of gender diversity in technical teams, while a better balance is achieved in product and sales teams. African Americans and Hispanics make up a lower share of the workforce. AI as their share in the general workforce, with the most severe decline coming from technical teams. These teams also have the highest proportion of Asian workers. “Interestingly, globally,” nearly 30% of scientific research papers in India include female authors, compared to an average of 15% in the US and UK, and well over 4% in China â.
Venture capitalists are reporting concerns about the management of big data in the AI ââspace. âCareful selection of data saves time and money by alleviating the challenges of big data. Working with large data sets is tedious and expensive. -ladder. Recent approaches address these computational costs, allowing the selection of data on modern datasets. “
Benaich and Hogarth highlight the need for better data, especially in real-time situations, such as the detection or prediction of life-threatening events. For example, they cite the threat of âdata cascadesâ, defined by Google researchers as âaggravating events causing negative downstream effects from data problemsâ. These researchers warn “that current practices underestimate data quality and lead to data cascades, pointing to factors such as” lack of recognition of data work in AI, lack of adequate training, difficulty in ‘access to specialized data for the region / population studied’. “This calls for” developing metrics to assess data quality, better incentives for data excellence, better data education, better practices for early detection of data cascades and better data access. ”
The VCs also anticipate that the coming year could see the launch of a General Artificial Intelligence (AGI) focused research company, “formed with strong support and a roadmap focused on a vertical industry, which could potentially involve development tools or a life science application.