For the general public, if they hear ‘great minds in ’, Elon Musk and Mark Zuckerberg are among the first to be mentioned. Of course, there have been many other great thinkers before them. One of them is Prof. Stuart Russell, whom SwissCognitive has had the pleasure of meeting multiple times in the past months.
Written by SwissCognitive
Stuart Russell has studied for all of his academic career and has been teaching at UC Berkeley since 1996. One year before, he published one of the standard works on “: A Modern Approach” together with Peter Norvig. Besides his teaching and researching responsibilities Stuart Russell is a great opponent of weaponized and has lent his voice to the short movie ‘Slaughterbots’.
Stuart Russell’s views on ’s current state, potential, and challenges are very insightful and thought provoking. Further, they illustrate what we as a community and society will have to continue to work on.
Leveraging data with
has brought the fourth industrial revolution along and we should all take advantage of it. As a matter of fact, to some extent has already been implemented in our everyday lives and we are already benefiting from it. It is still in its infancy stage however, but as we are dedicated to further improving our lives, advancing our technologies will remain our key tool in the process. The amount of data and knowledge that is available for us is enormous, and it needs to be put into use and further built on. While our current algorithms benefit from this massive amount of data, in the future we will have even better algorithms, which need fewer data points for better results.
Many routine tasks will be automatised, allowing us to focus on tasks which are more meaningful and more enjoyable for us. In fact, could help us to implement our human touch more where it is needed, and make use of automatisation and machine where humans cannot reach. There are many tasks that humans cannot do because we cannot process as much input as machines can, such as seeing the world as a whole or ploughing through all human knowledge in a feasible time frame. Nevertheless, there are tasks which require great contextual knowledge, as according to Stuart Russell’s example: the decision of whether or not the pet cat is a viable meat source for dinner. For us the answer is very clear, as we feel sentimental towards the cat, the however, might just see a great source of protein.
“If we are to build successful machines, they have to understand human preferences.” – Stuart Russell
Investing in good
The topic of has been astonishing us since decades. As many of its potentials have been only conceptualised but not yet necessarily taken advantage of, and as bad news tend to make it to the headlines more easily, the general public remains sceptical about its potentials. To realise the possibilities of , we need to understand, everything that exist today is a product of our intelligence. If we enable to have access to all the facts, data, intelligence, and human preferences that has ever existed and ever been recorded, we could multiply our productivity, efficiency, and success levels numerous times. Our decisions will be improved, and breakthrough findings will not only occur in our everyday lives, but also across industries, businesses, science and health care. The ultimate goal of applying is to make better decision – better than humankind has ever been capable of making.
Therefore, no wonder that the UK, France, China, the EU, and USA alone are investing tens of billions of dollars into . This is an investment put towards the future to ensure our life is becoming easier, our decisions better and our world more sustainable. The goal needs to be that implementing smart technologies becomes synonymous with the sentence “I am building a bridge”. No engineer has to specify that they are building a bridge ‘which will not fall down’, it is common sense. Similar to this, will have to always be provably beneficial for humans.
Quo vadis, ?
will change our lives tremendously. Change always brought and will always bring uncertainty and fear with it. Therefore, there is a lot of skepticism involved around . Will we all lose our jobs? How will we make sure that robots don’t take our world over? While progress has accelerated over the past decade, it is nearly impossible to make exact predictions on when we will have general , which is as or more intelligent as humans. Stuart Russell is among the scientists that believe that we will have software that is more intelligent as we are, the question is how we will prepare for it.
Researchers are exploring how we can reach the aforementioned provably beneficial . One idea is not to give software and the connected hardware a clear mission. We know from science-fiction movies like 2001 Space Odysee, that when Robots have a clear mission they will complete it without regards to collateral damage – whatever that may be. Provided that robots do not know that their mission is to serve humanity, but they are rewarded if ‘their’ human is prospering, the incentive to ensure the survival of humanity is immense. Of course, this might lead to a conflict of interest between the owner of the machine and everyone else. It might make me super happy, if ‘computer malfunction’ delays my flight, so I don’t miss it, however all the other passengers might not be as thrilled. Stuart Russell voiced an interesting thought, when he cautioned not to humanize robots too much, as it might be confusing to children, for example, if a nearly indistinguishable appears to be empathetic towards them, even though it is just a software. Furthermore, he said, if necessary, it is easier to cut the power from a grey box rather than a humanoid in case of a malfunction.
So far, autonomous vehicles are the most futuristic development in our lives. The aim is to make them as safe as possible as fast as possible, to avoid a public backlash and the consecutive set back in development. The software already has a much quicker reaction time, and therefore is able to ensure faster measurements to stop a moving vehicle. Moreover, it will allow for example elderly people to stay mobile for longer, while not endangering themselves and other traffic users, which has proven positive effects on their quality of live.
The future is ours to shape
We need to start working together, today, to build trust across societies and generations in the opportunities and benefits of . We need to take all the risks seriously and develop strategies which will help us to implement ethically, beneficially und sustainably. There is a great chance that – to our advantage and benefit – overtakes human abilities, and if we do it right, we have proven beneficial to augment our lives.
Let’s continue to connect on the topic of , share knowledge and experiences, debate about its development, as well as the benefits that it will bring along. SwissCognitive provides a community that does exactly that: a global hub, bringing experts and leaders around the table to reveal the potentials and challenges of , and share knowledge and experiences.
Image – Swiss Economic Forum