Organizations Struggle to Utilize Generative AI Safely, Reveals McKinsey Report
A recent report by McKinsey highlights a critical gap in preparedness among organizations for the safe and responsible deployment of generative AI. Although 40% of companies acknowledge the significant risks associated with AI explainability—the ability to comprehend how and why AI makes decisions—only 17% are taking proactive measures to address this issue.
In this landscape, Seoul-based startup Datumo is emerging as a beacon of hope for businesses striving to harness AI responsibly. Initially founded as an AI data labeling company, Datumo is evolving its mission to assist organizations in creating safer AI systems. On Monday, the startup announced a substantial funding round of $15.5 million, bringing its total funding to approximately $28 million, supported by a range of investors, including Salesforce Ventures and KB Investment.
Innovating for Accessibility
Datumo’s journey began with a frustration: CEO David Kim, a former AI researcher at Korea’s Agency for Defence Development, was disheartened by the tedious nature of data labeling. He conceptualized a reward-based app that allows anyone to contribute to data labeling and earn money in their free time. This innovative idea gained traction during a startup competition at the Korea Advanced Institute of Science and Technology (KAIST), where Kim co-founded Datumo in 2018 alongside five alumni from the institution.
Interestingly, even before the app was fully operational, Datumo generated tens of thousands of dollars in pre-contract sales during its customer discovery phase, primarily from businesses led by KAIST alumni.
From Data Labeling to AI Evaluation
Datumo quickly surpassed $1 million in revenue in its first year and landed contracts with prominent corporations, including Samsung, LG Electronics, and Hyundai. However, as demand evolved, clients began to request more than basic data labeling services. Recognizing this shift, Datumo pivoted towards AI model evaluation, releasing Korea’s first benchmark dataset focused on AI trust and safety.
"We started in data annotation, then expanded into pretraining datasets and evaluation as the LLM ecosystem matured," Kim noted.
Navigating Competitive Waters
Datumo’s strides in the AI sector reflect broader trends in the market. The recent $14.3 billion investment by Meta in data-labeling company Scale AI underscores the growing importance of this field. In a similar vein, Datumo aligns itself with established entities but sets itself apart through licensed datasets, sourced from published materials, offering unique structured human reasoning insights.
Moreover, Datumo’s full-stack evaluation platform, Datumo Eval, allows automated testing for unsafe or biased AI responses without requiring extensive technical expertise. This aligns with a biblical principle of stewardship—ensuring that technology serves to uplift and enhance the human experience, echoing the wise sayings of Proverbs 16:3: “Commit your work to the Lord, and your plans will be established.”
Looking Ahead
With the recent funding, Datumo aims to accelerate its research and development in automated evaluation tools and expand its market reach in South Korea, Japan, and the United States. This ambition resonates with the teachings of stewardship and diligence found throughout scripture, encouraging innovation that seeks the welfare of others.
As readers reflect on Datumo’s journey, consider the wider implications of using technology responsibly and ethically. In our fast-paced world, the invitation remains for us to be mindful stewards of the advancements we pursue, ensuring they align with principles of goodness, safety, and respect for one another.
Encouraging Takeaway: As you engage with technology and AI, think about how you can contribute positively to its development and application. How can your actions reflect principles of integrity, wisdom, and safety in this ever-evolving field?
Explore and dig up answers yourself with our BGodInspired App. Be careful – each interaction is like a new treasure hunt… you can get lost for hours 🙂
