Can Low‑Code Bridge the Gap Between Business and AI? Insights on Its Future

Invited by the Alibaba Cloud MVP team, the author attended an online developer conference and, driven by personal interests, shares thoughts on the future of low‑code, its relationship with algorithm development, and the evolving landscape of big data and AI.

The Future of Low‑Code

Low‑code is a hot yet controversial topic, often debated for its potential to replace traditional algorithm development. The author, experienced with Alibaba Cloud PAI and accustomed to Python and R, argues that low‑code and code can coexist, serving different audiences and use cases without one being inherently superior.

Algorithm design is both technical and artistic, producing models and predictions. Given the diversity of commercial and industrial scenarios and data types, algorithms will continue to evolve. Low‑code’s high‑level abstraction can limit flexibility, especially when granularity is too coarse, but overly fine granularity defeats its purpose and resembles traditional coding.

The key is to recognize low‑code’s strength: high encapsulation aimed at business users rather than developers or researchers. When algorithms are mature enough for production, low‑code shines as a bridge, enabling business personnel to participate in design and improvement, especially in industrial IoT where domain experts lack coding skills and developers lack domain knowledge.

Through modular assembly, business users can quickly test algorithm logic and receive immediate results. Feedback on suboptimal modules helps developers refine algorithms efficiently, reducing wasted effort and preserving valuable domain expertise within low‑code tools. Thus, low‑code’s greatest value lies in facilitating communication between specialties, accelerating the deployment of big‑data and AI solutions.

Fusion of Development and Algorithms

Insights from the conference highlighted improvements in data architecture and cloud‑based services. Development increasingly supports algorithms, and although they remain distinct fields, both aim to process data, necessitating integration to unlock data value.

Big data is often described by the 4V characteristics (volume, velocity, variety, veracity). Development addresses data variety by providing tools and databases for structured, semi‑structured, and unstructured data ingestion, while algorithms extract value from this data. Development also tackles data volume, leveraging cloud computing to provide the massive compute power required for machine learning and deep learning workloads.

Thus, development supplies the foundational infrastructure and security for algorithms, and robust compute resources enable large‑scale AI applications that were previously infeasible.

Future of Big Data and AI Data Processing

The future focus is on enhancing data cleaning capabilities. In many scenarios, the challenge is not data scarcity but data overload and low value density. Data cleaning can consume up to 70% of algorithm development effort and is expected to grow.

For AI, where data is often unstructured (audio, images, video), improved cleaning tools directly impact preprocessing, feature engineering, and overall model performance. As data volumes increase, the need for automated, practical cleaning solutions becomes critical.

Mid‑platform solutions aim to boost data processing and cleaning efficiency, which is arguably harder than scaling compute power. Advances in this area will transform raw data into valuable assets, complementing the growing compute capabilities showcased at the conference.