DeepSeek vs. ChatGPT: Which AI chatbot should you use?

Large Language Models (LLMs) like DeepSeek and ChatGPT are AI systems trained to understand and generate human-like text. Think of them as advanced assistants for writing, research, coding, or even brainstorming ideas.

DeepSeek, a Chinese AI startup, has introduced DeepSeek-R1, an open-source reasoning model designed to enhance problem-solving and analytical capabilities. OpenAI's ChatGPT is a well-established LLM known for its conversational abilities. But which one should you use for your daily musings? Can DeepSeek take over already or is ChatGPT still the king of the hill? Here's a quick comparison between the two AI chatbots.

Cost & Availability

Both chatbots have a free tier that allows users to ask their questions, get help with problem-solving, general information, etc. Both also have a paid subscription, with ChatGPT's models starting at $20/month, whereas DeepSeek starts at just $0.50/month, which is much more affordable.

DeepSeek-R1 is reportedly cost-effective to operate as well, with base fees approximately 27.4 times cheaper per token compared to OpenAI's o1.

Strengths & Weaknesses

DeepSeek

• Excels at generating code (Python, Java) and solving complex equations.
• Faces challenges on politically sensitive topics due to censorship protocols influenced by the Chinese government.

ChatGPT

• Better at storytelling, jokes, and marketing copy.
• More up-to-date on global events, and responds quicker when asked about recent events.
• Faces challenges related to generating contextually appropriate responses and mitigating biases inherent in its training data.

Real-World Testing

I am planning a 4-day trip to Singapore, so I asked both chatbots to create an itinerary for me. Here's the prompt I used: Plan a 4-day trip to Singapore for me

Both chatbots gave me fairly similar responses. However, ChatGPT went all out and suggested that I book a stay at some of the most luxurious hotels in Singapore. But I did like that at the end of the response, there was a section with estimated costs for the complete trip, with a suggestion asking me if I needed any adjustments to it.

DeepSeek-R1, on the other hand, gave me a slightly less detailed itinerary and didn't recommend any hotels. But for all my meals, the chatbot not only suggested restaurants but also recommended items on the menu that I could try, which was a nice touch. However, there was no mention of any budget or costs associated with the trip.

Since both chatbots also have the option to do a web search and get results based on real-time data, I tried another prompt: Tell me about Manchester City's recent form.

Do note that I did not select the option to turn on web searching on either chatbot. ChatGPT realised it on its own and gave me the most recent data, including the score from the latest football match played on January 25, along with inputs from publications like The Guardian and Reuters.

DeepSeek-R1 gave me an overview of Manchester City's recent form, but its data set cut-off was July 2024, which it promptly mentioned at the beginning of the response. So, I tried again and this time, I switched on the option to do a real-time web search. The chatbot took some time and eventually failed to respond, telling me that the demand was too high. 

"Oops! DeepSeek is experiencing high traffic at the moment. Please check back in a little while."

Ethics and Privacy

DeepSeek has committed to open-sourcing its models, promoting transparency and collaboration within the AI community. However, there have been concerns regarding potential alignment with governmental narratives, given its origin. It also refuses to answer sensitive questions related to China. I tried asking it about the relationship between Chinese President Xi Jinping and Russian President Vladimir Putin, and the chatbot immediately shut me down, telling me to "chat about math, coding, and logic problems instead!"

The same prompt on ChatGPT erupted into a response about how the duo share a partnership based on mutual interests. However, this is by no means a recuse for OpenAI, which has found itself in hot water over and over again. It is facing multiple copyright lawsuits in countries like India and USA. Elon Musk has also filed a lawsuit against OpenAI's leadership, including CEO Sam Altman, aiming to halt the company's transition to a for-profit model.

Which AI Chatbot Should You Use?

While ethics and privacy concerns are a large matter that both parent organisations will be dealing with, it isn't something that typically bothers the end consumer. So, which chatbot should you be using? There isn't one definitive answer, as both excel in areas where the others can be a step behind. 

You can use DeepSeek for 

• Queries related to math, logical reasoning or coding.
• More concise, technical answers with a thought process that shows how the chatbot got to the final output.

You can use ChatGPT if you

• Want creativity and conversational flair.
• Need up-to-date info on current events.

Neither tool is "better", it depends on your needs. DeepSeek shines for developers and students tackling technical tasks, while ChatGPT still remains the go-to for everyday users seeking engaging, human-like interactions.