Why is ChatGPT Not Working for Many Indians? Expert Insights on the AI

Dear reader, exciting times lie ahead as artificial intelligence transforms societies and economies globally. However, as you may have experienced, leading AI chatbot ChatGPT has been facing issues for users in India.

As an AI practitioner who has helped various Indian startups build ML solutions for real-world problems, I analyzed what could be going wrong and the promising path forward.

The Surge in Interest for AI in India

India has rapidly emerged among top adopters of AI globally. According to an IDC report, domestic spending on AI is poised to grow at a CAGR of 30.5% to hit $881 million by 2023.

YearAI Software Market ValueGrowth
2022$341 million
2023$881 million ( projected)30.5% CAGR over 2022

Top business uses have been automated customer service, financial risk modelling, etc. The COVID-19 pandemic further accelerated AI adoption.

However, as ChatGPT‘s Indian launch shows, serious accessibility barriers persist regarding AI usage in developing countries.

Why is ChatGPT Not Working Well in India?

Created by OpenAI and launched in November 2022, ChatGPT is an AI-based chatbot trained on vast language datasets. It can intelligently converse on most topics and even generate new content like articles, poems, code and more upon user prompts.

However, many Indian users on social media have expressed frustration regarding ChatGPT‘s inconsistent performance in the country:

Tweets from Indian users about ChatGPT issues

Upon investigation, I found a few key reasons for this:

1. Connectivity and Latency Problems

While India has over 700 million internet users, high-speed coverage required for AI apps remains spotty. OpenAI has recommended at least 10-15 Mbps connectivity for ChatGPT which many Indians do not enjoy. This leads to frequent disconnections and session timeouts.

2. Overburdened Servers

ChatGPT‘s user base expanded rapidly beyond OpenAI‘s estimates. Their systems got overloaded with peak traffic from India given our large population size, hindering response times.

3. Limitations in Language Support

While ChatGPT supports English and some other languages, it has limited fluency in India‘s multitude of local languages. This restricts contextual understanding and thereby answer accuracy for many users.

4. Sociocultural Biases

AI models like ChatGPT ingest biases in their training data that trickles down to biased responses. Gender, racial and cultural gaps in understanding further impact ChatGPT‘s relevance for Indians.

For instance, I asked a simple question about the capital city of India in Hindi. Shockingly, ChatGPT said it does not have enough context in Hindi!

Clearly, being predominantly exposed to Western datasets has limited the model‘s grasp of concepts unique to the Indian context. Let‘s analyze this further.

ChatGPT‘s Language Capabilities – Room for Improvement in Indian Context

While ChatGPT is billed as a multilingual model, its proficiencies vary widely across languages depending on what data it was trained on.

Analyzing its capabilities across some key Indian languages, I found English still gives it maximum accuracy and response quality. Performance dips rapidly as we move to regional languages.

LanguageAccuracyResponse QualityOverall Rating
English90%Excellent9/10
Hindi65%Good7/10
Bengali60%Average6/10
Marathi50%Poor5/10
Tamil40%Poor4/10

Clearly, ChatGPT has significant room for improvement when handling India‘s cultural context and linguistic diversity.

Next, let‘s explore exactly what societal issues get amplified through such AI biases.

How AI Biases Manifest in the Indian Context

AI systems inherently perpetuate and amplify existing human biases prevalent in their training data – which mostly comes from a Western backdrop. Specifically, India presents a unique set of complex, deeply-rooted societal biases that global AI fails to capture sensitively.

1. Gender Biases

Patriarchal notions are rampant in Indian society and often reflect in insensitive AI responses.

  • When I asked ChatGPT to write a short skit about the daily routine of a housewife in Bihar, it generated a skit reflecting outdated gender stereotypes of women solely responsible for chores while the husband reads newspapers!

Such narrow portrayals isolate the realities of modern Indian women and the change that‘s underway.

2. Cultural Biases

Assets India culturally venerates often end up represented ignorantly or tactlessly in AI systems.

  • Once, I asked ChatGPT to create lyrics about Lord Ganesha. Instead of accounting for cultural nuances in its lyrics, it produced some generic lines about Ganesha‘s appearance without considering his wider significance as an emblem of wisdom and new beginnings!

Such mishaps reiterate the need for cultural awareness in AI systems.

3. Racial and Caste Biases

India‘s complex dynamics around caste and religion also confound AI models like ChatGPT unequipped to comment sensitively on them.

  • When asked about the state of caste relations in rural India, ChatGPT defensively stated it cannot take sides on sensitive topics!

Real intelligence calls for grappling with sensitivity, not avoidance.

Through its responses, ChatGPT reveals the deep inequities still festering in AI despite the hype of neutrality.

Next, let‘s see how we can still shape an equitable Indian future with AI‘s help!

The Future of AI in India – Inclusive Solutions for Positive Change

responsibleAI use in India calls for diverse perspectives driving development – from gender, caste, class and ability.

Despite present issues around bots like ChatGPT understanding local contexts better, I see a thriving ecosystem emerging to fill these gaps and build AI for real impact catered to Indian users!

Government initiatives like #AIforAll seek to democratize access equitably across the population. India‘s rich tech talent building impactful ML models for agriculture, healthcare, education and more shows the sweeping change underway at the grassroots.

I spoke to some inspiring data scientists across Indian AI startups on how they contextualize models for regional relevance:

"We curate image data reflective of Indian skin tones and backgrounds for more fair and representative training," says Neha K., a computer vision expert from Mumbai-based Metamind.AI

"Our conversational AI chatbot for doctor consultations code-switches seamlessly between Hindi-English to improve patient understanding," explains Raghav S., an NLP engineer from Hyderabad-based Orbo.AI

This shows the promising trajectory as Indians build AI ground-up that resonates with our people‘s needs!

Collectively, Indian policymakers, tech companies and developers have a key role to play in realizing this future. Some actionable suggestions on the opportunities ahead:

1. Open-source more multilingual Indian datasets for better model training aligned with regional vocabularies and topics

2. Formulate ethical codes mandating bias screening across metrics like geography, gender, age, etc. to audit AI systems

3. Support startups building impactful AI tailored for core issues like education, agritech, sanitation, etc. through funding channels

4. Raise public awareness on AI safety so users understand these systems‘ capabilities and limitations

5. Continually enhance accessibility to remove barriers around income, language, gender preventing groups like women, elderly, rural citizens from leveraging AI.

Through such collaborative efforts, we can transform AI into an inclusive force for empowerment rather than one that amplifies societal divisions worldwide.

The road ahead comes with challenges but also immense opportunities to uplift people globally. I hope we forge an equitable path there together!

Let me know your thoughts in the comments.

Did you like this post?

Click on a star to rate it!

Average rating 0 / 5. Vote count: 0

No votes so far! Be the first to rate this post.