1 week ago
Andhra Pradesh AI GST Model Draws National Attention
Andhra Pradesh showed other Indian tax officials a computer system that helps manage GST work.
The system uses artificial intelligence and machine learning to find cases that may need closer attention.
It studies information from four sources and ranks cases using 35 risk factors.
Another tool helps officials prepare legal documents using tax laws and about 22,000 court decisions.
It has helped with more than 13,700 cases.
During six months, the AI-supported process detected more possible tax revenue than the older manual process.
The computer suggests information and drafts, but government officers still make and sign the final decisions.
Other states were interested and may visit Andhra Pradesh to learn how the system works.
Andhra Pradesh presented its AI- and ML-based GST administration model at the sixth National GST Coordination Meeting in New Delhi.
The system combines data from four sources, 19 analytical reports and a 35-parameter risk matrix to classify cases.
Its Legal-AI Officer Assistant, trained on GST law and about 22,000 judicial precedents, has supported more than 13,700 cases.
AI-assisted return scrutiny detected Rs 743.43 crore over six months, compared with Rs 365.75 crore under the earlier manual system.
Several states requested system details and plan study visits, while Andhra Pradesh proposed expanding the model to refunds, registrations, recoveries and advance rulings.
- Who
- Andhra Pradesh’s Commercial Taxes Department, represented by chief commissioner of state tax Babu Ahmed, presented the model to central and state tax officials.
- What
- It showcased an AI- and ML-based system covering GST case selection, scrutiny, audits, inspections and litigation support.
- Where
- Vigyan Bhawan in New Delhi.
- When
- At the sixth National GST Coordination Meeting on Friday; the articles do not provide a calendar date.
- Why
- To improve GST administration, identify high-value cases, support legal work and offer a model that other states could adopt.
Key facts
- Meeting
- Sixth National GST Coordination Meeting in New Delhi
- Case-classification system
- Four data sources, 19 analytical reports and a 35-parameter risk matrix
- Legal-AI training
- GST statutory material and about 22,000 judgments from courts and tribunals
- Cases supported
- More than 13,700 cases
- AI-assisted detection
- Rs 743.43 crore during six months of AI-driven return scrutiny
- Manual comparison
- Rs 365.75 crore during the comparable earlier manual period
- Detection per case
- Rs 27.63 lakh with AI assistance versus Rs 3.08 lakh under the earlier process
Quotes
Andhra Pradesh Commercial Taxes Department
The state tax department describing how automation remains subject to statutory officer accountability
“AI drafts, but officers decide”
deccanchronicle.com










