You should never ever outsource anything to AI models
Join our daily and weekly emails to receive the latest news and exclusive content about industry-leading AI. In a world of disruption and efficiency, businesses will look to generative AI for a powerful ally. OpenAI’s chatGPT, which generates human-like texts, and DALL-E, which produces art on demand, have shown us glimpses of the future, where machines will create with us, or even lead. Why not apply this to research and development? AI can boost idea generation and iteration faster than humans, and could even discover the next big thing with astounding ease. It sounds good in theory, right? But let’s be honest: relying on Gen AI to do your R&D is likely to backfire in major, and even catastrophic ways. Outsourcing generative tasks is dangerous, whether you are an established company defending your turf or a startup looking to grow. Let me explain why an over-reliance on Gen AI in R&D may be innovation’s Achilles heel. The unoriginal genius of AI: Prediction
imagination
Gen AI is essentially a supercharged prediction machine. It predicts what words, designs, code or images will work best, based on an extensive history of previous examples. Let’s not be deceived: AI is only good as the data it has. It is not creative in the sense that we understand it; it does not “think” in a radical or disruptive way. It is always looking backwards, relying on the past. For you to truly create something new, it takes more than incremental improvements extrapolated from past data. Great innovations are often the result of leaps, pivots and reimaginings. Not a minor variation on a theme. Apple, with its iPhone, or Tesla in the electric car space did not just improve existing products; they turned paradigms upside down. Human imagination is what creates the bold, life-changing moments that change markets, behaviors and even industries. Not probabilities calculated by algorithms. When AI is driving your R&D, you end up with better iterations of existing ideas, not the next category-defining breakthrough.
2. Gen AI has a homogenizing nature.
Letting AI drive your product ideation can be a dangerous move. AI will process content, whether it’s designs, solutions or configurations on telecommunications equipment in ways that encourage convergence instead of divergence. AI-driven R&D results in homogenized products due to the overlap of training data. Yes, different flavors of the same concept, but still the same concept.Imagine this: Four of your competitors implement gen AI systems to design their phones’ user interfaces (UIs). Each system is taught on the same information – data scraped off the internet about consumer preferences, current designs, bestsellers, etc. What are the results of all these AI systems? What you will see is variations of the same result.Over time, you’ll notice a disturbing cohesion in visual and conceptual design where rival products begin to mirror one another. The icons may be slightly different or the features of the products will vary at the margins but what about substance, uniqueness and identity? Pretty soon, they evaporate.
We’ve already seen early signs of this phenomenon in AI-generated art. Many artists on platforms such as ArtStation have expressed concern about the AI-produced content, which, rather than showing human creativity, feels recycled, remixing broad visual tropes, popular cultural references and styles. If every company uses gen AI to drive innovation, your industry will not get five or 10 disruptive products per year. Instead, it will get dressed-up copies. The magic of human mischief: How accidents and ambiguity propel innovation
We’ve all read the history books: Penicillin was discovered by accident after Alexander Fleming left some bacteria cultures uncovered. Percy Spencer, an engineer at the time, accidentally melted chocolate by standing too near a radar. And the Post-it? A failed attempt to create a super-strong glue was another happy accident. Human researchers are uniquely able to recognize the hidden value in failure and can often see the unexpected as an opportunity. Serendipity and intuition are just as important to innovation as a carefully-planned roadmap. Its programming is designed to teach it how to avoid errors, optimize accuracy, and resolve data ambiguities. That’s great if you’re streamlining logistics or increasing factory throughput, but it’s terrible for breakthrough exploration.
By eliminating the possibility of productive ambiguity — interpreting accidents, pushing against flawed designs — AI flattens potential pathways toward innovation. Humans are able to embrace complexity, and they know how breathe when unexpected results occur. AI will, on the other hand, double down in its certainty and mainstream middle-of-the-road ideas, while ignoring anything that appears irregular or untested. AI lacks empathy and vision — two intangibles that make products revolutionary
Here’s the thing: Innovation is not just a product of logic; it’s a product of empathy, intuition, desire, and vision. Humans innovate not only to respond to logical efficiency, but also human emotions and needs. We dream of making things faster, safer, more delightful, because at a fundamental level, we understand the human experience.
Think about the genius behind the first iPod or the minimalist interface design of Google Search. These game-changers were not only successful because of their technical merit, but also because they had the empathy to understand users’ frustrations with complicated MP3 players and cluttered search engines. This is something that Gen AI can’t replicate. It can’t understand what it is like to struggle with a buggy application, marvel at an elegant design, or feel frustrated by a need that hasn’t been met. AI does not “innovate” with emotional context. Its lack of vision makes it less able to create points of view which resonate with real humans. Without empathy, AI can create products that may be technically impressive, but lack humanity, feel sterile, transactional, and soulless. This is a killer for innovation in R&D. To much dependence on AI can de-skill human talent.
Here is a final thought that will chill our AI-future fans. What happens if you allow AI to do too much work? Skills degrade in any field where automation reduces the level of human involvement. Look at industries that have been heavily automated: Employees are losing touch with “why” things happen because they don’t exercise their problem-solving skills regularly. Research teams that become passive overseers of AI-generated output may lose their ability to challenge, outthink, or surpass the AI’s work. You will become less innovative the less you innovate. This erosion of human skills is dangerous in times of market change, when AI cannot guide you through the fog. AI is not suited to the challenges of disruptive times. Humans must be able to think outside the box. AI is a powerful tool that can help researchers and designers test hypotheses, refine ideas and iterate quickly. Used properly, it can enhance productivity without squashing creativity.
The trick is this: We must ensure that AI acts as a supplement, not a substitute, to human creativity. Human researchers need to stay at the center of the innovation process, using AI tools to enrich their efforts — but never abdicating control of creativity, vision or strategic direction to an algorithm.
Gen AI has arrived, but so too has the continued need for that rare, powerful spark of human curiosity and audacity — the kind that can never be reduced to a machine-learning model. Ashish Pawar, a software developer, is an expert in this field.
DataDecisionMakers
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!
Read More From DataDecisionMakers
