Alex Wong Cian Yih OP Deltaman099 : I understand your perspective, and I think it’s an interesting point. But let me ask you: if you were running OpenAI and had invested billions of dollars in training models using massive compute resources and human feedback, only to see a competitor use your model’s outputs to build their own product at just 1/20th of your cost—how would you feel? Do you think it would still motivate you to invest in groundbreaking innovation, knowing your efforts could be leveraged by others so easily? I’m genuinely curious to hear how you’d approach this situation if the roles were reversed.

Alex Wong Cian Yih OP lousyimpressario : I see your point, but there’s a fundamental difference between drawing inspiration from public knowledge and directly leveraging a competitor’s proprietary outputs to train a competing model.
OpenAI’s API policy explicitly states that its outputs cannot be used as training data to develop competing AI models. This is clearly outlined in their Terms and Conditions (T&C), and DeepSeek has blatantly ignored this rule. OpenAI’s API-generated responses are not just raw data scraped from the internet—they are already fully processed, cleaned, and structured high-quality outputs, carefully curated through human feedback and extensive model training.
What this means is that DeepSeek didn’t have to go through the same rigorous and expensive data processing steps as OpenAI, such as collecting raw internet data, performing human annotation, conducting reinforcement learning with human feedback (RLHF), and filtering out low-quality responses. Instead, they simply took OpenAI’s high-quality API outputs and fed them back into their own model, effectively short-cutting the most costly and labor-intensive parts of AI training.
This is why, when asked about its origins, DeepSeek’s model has been seen stating that it was trained by OpenAI—a clear indication that it directly absorbed OpenAI’s API responses without meaningful transformation.
Yes, knowledge distillation and model refinement are common in the industry, and many AI companies do use competitive insights to improve their models. However, other companies at least process, refine, and adjust the data before using it, ensuring that their outputs are not direct copies. What DeepSeek has done is take OpenAI’s processed knowledge as-is and repurpose it for its own use without adding real value or originality.
This isn’t just an ethical concern—it’s a direct violation of OpenAI’s T&C and an unfair exploitation of their proprietary research. If this practice were normalized, it would undermine the incentives for any company to invest in real AI innovation, as any competitor could simply take the end product and reuse it for free. OpenAI, and any other serious AI developer, would have little reason to continue advancing the field if their outputs could be so easily repurposed by rivals.
So, this is not just about ‘inspiration’ or ‘competition’—it’s a clear-cut case of unauthorized exploitation of another company’s intellectual property, and it sets a dangerous precedent for the future of AI development.

Alex Wong Cian Yih OP lousyimpressario : I see what you’re saying, but I think your argument overlooks some critical distinctions. Innovation is indeed about building on prior knowledge, but there’s a major difference between drawing from public data and directly leveraging a competitor’s proprietary outputs.
Your chef analogy doesn’t work because OpenAI’s API responses aren’t just ‘flavors to be tasted’—they’re more like a secret recipe that took billions to perfect. DeepSeek isn’t just drawing inspiration from OpenAI; they’re accused of directly training on its proprietary outputs, which OpenAI’s Terms & Conditions explicitly prohibit.
You also assume this won’t impact AI investment, but that’s an overly optimistic view. If major AI developers can’t protect their research, they will have little incentive to keep pushing boundaries. Why spend billions when a competitor can just extract the final results and skip the hardest steps?
Fair competition should be about genuine innovation, not bypassing a competitor’s most difficult and costly R&D efforts. That’s the real issue here.

Alex Wong Cian Yih OP Deltaman099 : I find it interesting that you immediately default to 'just take it to court' when we are simply discussing an industry-wide issue. I’m not OpenAI, nor am I an investor in OpenAI—so why would legal action be my concern? What we are debating here isn’t about filing lawsuits; it’s about ethics, fair competition, and the impact on AI innovation as a whole.
Legal action isn’t the only way to discuss ethical concerns. Even before lawsuits happen, unfair practices can impact the AI industry and competitive landscape. The real question isn’t whether DeepSeek optimized their model architecture, but whether they followed OpenAI’s terms of service when acquiring training data.
It doesn’t matter how well a model performs if its foundation is built on violating another company’s policies. Innovation should be fair, and genuine breakthroughs should come from original work—not from repurposing a competitor’s proprietary outputs without permission.
If we can’t even discuss these concerns rationally, then by your logic, we shouldn’t question any potential violations in any industry unless a court case is filed. That’s not how ethical discussions work.

Alex Wong Cian Yih OP Deltaman099 : Since DeepSeek has open-sourced its model under the MIT license, OpenAI can, in theory, apply knowledge distillation techniques to improve its own models using DeepSeek’s outputs. MIT licensing allows commercial and unrestricted use, meaning OpenAI could legally extract insights from DeepSeek’s model to enhance its Chinese-language capabilities.
Given that DeepSeek’s advancements in Chinese comprehension, grammar, and semantics are notably strong, leveraging its outputs could significantly enhance OpenAI’s own models in this area. This is a practical example of how open-source fosters cross-learning—just as DeepSeek might have built upon existing AI advancements, OpenAI now has the same opportunity to do so in return.
With DeepSeek’s model available to everyone, it becomes part of the broader AI ecosystem, allowing improvements to flow in multiple directions. This is how technological progress works when models are shared openly.

Alex Wong Cian Yih OP bullbearnme : Thank you for your feedback. I have over five years of experience working in data centers. If you're interested in verifying my background, please feel free to check my LinkedIn profile.

Alex Wong Cian Yih OP bullbearnme : Thank you for sharing these market projections. While they highlight the potential growth of quantum computing in cloud data centers, they are based on several assumptions. Quantum computing still faces significant technical hurdles, especially in scaling and error correction for large deployments. Additionally, its primary applications are in specialized fields like drug development, earthquake prediction, protein folding, and new material research—areas that most small and medium-sized enterprises typically do not require. Many quantum startups may also struggle to sustain themselves until these forecasts materialize.

Alex Wong Cian Yih OP Agent Smith 404 : I think there’s been a misunderstanding. My article wasn’t dismissing XRP’s potential but emphasizing the importance of managing risks and locking in profits during volatile times. While XRP’s recent rise is impressive, surpassing Bitcoin and Ethereum in market cap is a massive challenge based on current market dynamics. I respect your enthusiasm, but my focus is on realistic expectations and responsible investing.

Alex Wong Cian Yih OP Mooge : Your concern is valid, and there are indeed reports suggesting that Apple's plan to produce its own 5G modem processors might be delayed until 2026, rather than 2025, which means their partnership with Qualcomm could extend further than initially expected. Additionally, while Apple contributes to Qualcomm's revenue through its 5G chips, it represents only a portion of their overall business. Qualcomm has been diversifying into other areas like IoT and automotive, which will gradually offset any potential loss. The impact, though present, is unlikely to be significant enough to warrant major concern at this point.