Microsoft is back to its winning ways. And with its surging profits, the company is looking for ways to marshal its enormous resources to keep up the momentum. Perhaps one of the most important recent deals is a $1 billion investment in OpenAI. The goal is to build a next-generation AI platform that is not only powerful but ethical and trustworthy
Founded in 2015, OpenAI is one of the few companies — along with Google’s DeepMind — that is focused on AGI (Artificial General Intelligence), which is really the Holy Grail of the AI world. According to a blog post from the company: “AGI will be a system capable of mastering a field of study to the world-expert level, and mastering more fields than any one human-like a tool which combines the skills of Curie, Turing, and Bach. An AGI working on a problem would be able to see connections across disciplines that no human could. We want AGI to work with people to solve currently intractable multi-disciplinary problems, including global challenges such as climate change, affordable and high-quality healthcare, and personalized education. We think its impact should be to give everyone economic freedom to pursue what they find most fulfilling, creating new opportunities for all of our lives that are unimaginable today.”
It’s a bold vision. But OpenAI has already made major advances in AI. “It has pushed the limits of what AI can achieve in many fields, but there are two, in particular, that stands out,” said Stephane Rion, who is the Senior Deep Learning Specialist of Emerging Practices at Teradata. “The first one is reinforcement learning, where OpenAI has driven some major research breakthroughs, including designing an AI system capable of defeating most of its human challengers in the video game Dota 2. This project doesn’t just show the promise of AI in the video game industry, but how Reinforcement Learning can be used for numerous other applications such as robotics, retail and manufacturing. OpenAI has also made some major advances in the area of Natural Language Processing (NLP), specifically in unsupervised learning and attention mechanism. This can be used to build systems that achieve many language-based tasks such as translation, summarization and generation of coherent paragraphs of text.”
But keep in mind that AI is still fairly weak — with applications for narrow use cases. The fact is that AGI is likely something that is years away. “In fact, we’re far from machines learning basic things about the world in the same way animals can,” said Krishna Gade, who is the CEO and Co-founder at Fiddler Labs. “It is true, machines can beat humans on some tasks by processing tons of data at scale. However, as humans, we don’t understand or approach the world this way — by processing massive amounts of labeled data. Instead, we use reasoning and predictions to infer the future from available information. We’re able to fill in gaps, extrapolate things with common sense and work with incomplete premises — which machines simply can’t do yet. So while it is interesting, I do believe that we’re quite far from AGI.”
Guy Caspi, who is the CEO of Deep Instinct, agrees on this. “While deep learning has drastically improved the state-of-the-art in many fields, we have seen little progress in AGI,” he said. “A deep learning model trained for computer vision cannot suddenly learn to understand Japanese, and vice versa.”
Regardless, with OpenAI and Microsoft willing to take on tough challenges, there will likely be an acceleration of innovation and breakthroughs — providing benefits in the near-term. “I love to see a company like Microsoft, which the market has rewarded with enormous returns on capital, making investments that aim to benefit society,” said Dave Costenaro, who is the head of artificial intelligence R&D at Jane.ai. “Funding OpenAI, creating more open source content and sponsoring their ‘AI for Earth’ grants are all examples that Microsoft has recently spun up in earnest. Microsoft is not a charity, however, so the smart money will take this as a signal of how strategically important these issues are.”
What’s more, the emphasis on ethics will also be impactful. As AI becomes pervasive, there needs to be more attention to the social implications. We are already seeing problems, such as with bias and deepfakes.
“With the fast pace of AI innovation, we’re continually encountering new, largely unforeseen ethical implications,” said Alexandr Wang, who is the CEO of Scale. “In the absence of traditional governing bodies, creating ethical guidelines for AI is both a shared and a perpetual responsibility.”