Appen, an artificial intelligence (AI) assisted data annotation platform with 1125 employees globally, recently published a whitepaper entitled The State of AI and Machine Learning. It wasn’t really a comprehensive review of artificial intelligence changes but it did touch on a few interesting points about “How the AI industry continues to evolved in a world that is more virtual, more tech-savvy and more globalized than ever.”
Appen found that artificial intelligence budgets are predominantly in the $500K to $5M range. Budgets over $500K increased from 34% of the total in 2020 to 53% in 2021. Budgets under $500K went from 40% of the total in 2020 to 26% in 2021. In total, budgets from $500K to $5M have increased year-over-year by 55%. Appen says, “This trend is a strong signal that the industry continues to grow and AI is becoming more critical to the success of companies large and small across all industries.”
Instead of the c-suite focusing on the need for AI in the organization, responsibility for AI projects has become more operational. C-level executives were responsible for 71% of AI initiatives in 2020 but that dropped to 39% in 2021. Large companies are delegating AI responsibility to the VP level (28%) and Director level (25%) while smaller organizations are seeing managers (31%) take on the artificial intelligence changes.
In 2020 80% of organizations updated their AI models at least quarterly with 45% doing it monthly. In 2021 87% updated their models quarterly with 57% doing it monthly. “It is critical to update and retrain AI models to have a successful model and avoid data drift. 87% of organizations are updating their models regularly with 91% of large organizations updating at least once a quarter. When data is always changing and use cases are evolving, this becomes a big part of maintaining the accuracy of your model,” says Wilson Pang, Appen Chief Technical Officer.
Obtaining Training Data
“Obtaining sufficient high-quality training data to deploy AI is a significant barrier to success for organizations of all sizes,” says the report of a company providing training data. They do support this statement though with some relevant statistics. Their study found that 90% of small companies, 92% of medium companies and 85% of large companies use external providers of data. It is true that data acquisition, preparation and management are top challenges in the effective use of ai.