Category Archives: Ai

Nvidia said it plans to train 100,000 developers through its Deep Learning Institute.

For Nvidia, the Deep Learning Institute, an effort to train developers in machine learning and artificial intelligence, is a way to create a well of expertise that can ultimately lead to more sales of GPUs.

The bet for Nvidia is that IDC estimates that 80 percent of all applications will have AI as a component by 2020.

Nvidia's Deep Learning Institute launched a year ago and has held training events at academic institutions, companies and government agencies. So far, Nvidia's efforts have trained more than 10,000 developers who use Amazon Web Services (AWS) EC2 P2 GPU instances.

How to Implement AI and Machine Learning

The next wave of IT innovation will be powered by artificial intelligence and machine learning. We look at the ways companies can take advantage of it and how to get started.

Greg Estes, vice president of developer programs at Nvidia, acknowledged that training 100,000 developers in 2017 is ambitious, but added that there is strong demand and expanded content can broaden the audience.

In an effort to train 100,000 developers in the next year, Nvidia has stepped up its offerings with the following:

Estes told journalists at Nvidia GTC it made sense for the company to partner with larger companies.

"They are going to help us expand our reach ... because these companies are much bigger than we are, and they have a lot of worldwide reach," he said.

"I think most people would agree that we are at the very leading edge of artificial intelligence and deep learning -- so if we take our knowledge and expertise there, and we work with these other companies, they can help bring that out into the community -- it's a win for everybody."

In the coming year, Nvidia is also planning to certify engineer competence.

"Today when you go through and you take these learning courses, we give you a certificate that you have attended the course, but we don't have the testing at the end," he said. "That is on our roadmap, and we plan to do that this year."

See more here:

Nvidia aims to train 100000 developers in deep learning, AI technologies - ZDNet

According to NVIDIA, there are already hundreds of millions of surveillance cameras around the globe, with the number expected to rise to the 1 billion mark by 2020. Human beings have a hard time sifting through the flood of moving images, storing the majority of it on hard drives for later viewing. NVIDA thinks that deep learning AI can help video analytics much more accurately than humans or even real-time computer monitoring. The company has partnered with more than 50 companies that make security cameras, including Hikvision. "The benefit of GPU deep learning is that data can be analyzed quickly and accurately to drive deeper insights," said Shiliang Pu, president at the Hikvision Research Institute in China.

A city with cloud-connected, AI-powered surveillance systems in place could find missing persons, notify residents of nearby emergencies, alert police to crimes in progress or even send out traffic congestion warnings. It could also track and monitor our behavior both legal and otherwise along with gathering personal data for advertisers. Tomorrow can be both exciting and scary at the same time. Whether the city of the future keeps us safe, keeps us in line or something in-between will depend on how we implement emerging technology like this now.

Go here to see the original:

NVIDIA's AI may keep watch over smart cities of the future - Engadget

Baidu, Inc. (NASDAQ: BIDU) is the largest online search engine in China, and since its entry into the realm of artificial intelligence (AI), the company has been integrating its AI know-how into nearly every facet of its business. What investors may not know is that this process produces a virtuous cycle that continues to feed itself.

In the case of Baidu, it has been engaged in an area of AI known as deep learning. Algorithms and software models are used to develop artificial neural networks that mimic the human brain's ability to learn. Vast amounts of data are required to train the system, and Baidu's online search engine provides a vast depository of information on which to draw. Once trained, these AI systems are then used to process data at much faster rates than their human counterparts can and are skilled at detecting patterns. One key aspect of deep learning is that these systems become more useful the more they are used.

Baidu is in a virtuous cycle with its AI. Image source: Baidu.

The ability to detect patterns can be used in a wide array of areas, and these systems are particularly skilled at tasks such as image recognition, making more precise online search recommendations, and more accurately predicting traffic conditions for users of its Maps service.

The company has also applied its AI acumen to better estimating delivery times for its Baidu Delivery service and making customized restaurant suggestions for its recommendation platform, Nuomi. It uses AI to recommend content to its millions of users of its iQiyi video streaming service and provide more relevant content for its news feed.

The process of improving products with AI is a two-way street. When it makes recommendations, the AI system receives feedback from these diners and drivers and streamers, which allows the system to make more reliable recommendations in the future. The system improves with each interaction.

Baidu is using its AI system to develop self-driving car technology in China. It recently announced the acquisition of xPerception, a start-up in the field of computer vision. The company focuses on object recognition and depth perception that can be used in the area of autonomous vehicles and can also be used for drones.

This virtuous cycle is the secret sauce of Baidu's AI system. By using AI, it improves its products and recommendations. By integrating the feedback into the AI system, it improves the relevance of future recommendations. This holds true across the plethora of ways that Baidu is using its AI.

Baidu also announced that it would open-source its platform for autonomous driving, in a move that was said to be inspired by Google's open sourcing of its Android platform. The Alphabet Inc. (NASDAQ: GOOGL) (NASDAQ: GOOG) division came to dominate the smartphone market with its Android operating system, by making it available to all comers, thereby maintaining its supremacy in mobile search.

This is a smart move, as the cumulative data that's acquired from self-driving cars will make the systems safer, and Baidu hopes to become the de facto leader of such data in its native China.

Baidu's strategy regarding AI has yet to bear fruit. While the company had been investing large sums to bolster its AI capability, and it has been seeing incremental improvements in a variety of areas, it has yet to produce any significant increase in revenue.

In the most recent quarter, the company marked its third consecutive quarter of declining revenue, though Baidu hopes to return to growth later this year. The revenue from the company's most recent quarter increased 6.8% over the previous year quarter to $2.45 billion, and earnings of $258 million were down by 10.6% year-over-year.

The area of artificial intelligence offers exciting possibilities, but investors would more likely be excited by a return to growth in revenue and earnings.

10 stocks we like better than Baidu

When investing geniuses David and Tom Gardner have a stock tip, it can pay to listen. After all, the newsletter they have run for over a decade, Motley Fool Stock Advisor, has tripled the market.*

David and Tom just revealed what they believe are the 10 best stocks for investors to buy right now... and Baidu wasn't one of them! That's right -- they think these 10 stocks are even better buys.

*Stock Advisor returns as of May 1, 2017

Suzanne Frey, an executive at Alphabet, is a member of The Motley Fool's board of directors. Danny Vena owns shares of Alphabet (A shares) and Baidu. Danny Vena has the following options: long January 2018 $640 calls on Alphabet (C shares) and short January 2018 $650 calls on Alphabet (C shares). The Motley Fool owns shares of and recommends Alphabet (A shares), Alphabet (C shares), and Baidu. The Motley Fool has a disclosure policy.

Visit link:

Baidu is Using AI to Improve Its Products -- and Its Products to Improve Its AI - Madison.com

Microsoft wants AI to help, not replace humans.

It's been abundantly clear for a while that in 2017, artificial intelligence (AI) is going to be front and center of vendor and enterprise interest. Not that AI is new - it's been around for decades as a computer science discipline. What's different now is that advances in technology have made it possible for companies ranging from search engine providers to camera and smartphone manufacturers to deliver AI-enabled products and services, many of which have become an integral part of many people's daily lives.

More than that, those same AI techniques and building blocks are increasingly available for enterprises to leverage in their own products and services without needing to bring on board AI experts, a breed that's rare and expensive.

How to Implement AI and Machine Learning

The next wave of IT innovation will be powered by artificial intelligence and machine learning. We look at the ways companies can take advantage of it and how to get started.

Sentient systems capable of true cognition remain a dream for the future. But AI today can help organizations transform everything from operations to the customer experience. The winners will be those who not only understand the true potential of AI but are also keenly aware of what's needed to deploy a performant AI-based system that minimizes rather than creates risk and doesn't result in unflattering headlines.

These are the three key challenges all AI projects must tackle:

Interested in a deeper dive? I'll be covering this topic at Forrester's Digital Transformation Europe Forum in London on June 8-9. Click here to register today.

Martha Bennett is principal analyst at Forrester. Follow Martha on Twitter: @martha_bennett.

Here is the original post:

Three key challenges that could derail your AI project - ZDNet

The term artificial intelligence (AI) has been around since John McCarthy invented it in 1955, describing the field as "the science and engineering of making intelligent machines. Yet today, the topic is increasingly given overwhelming attention in the advertising industry as though it was thought of in recent years.

A new tech product with the incorporation of AI is thought to be more advanced and revolutionary. But is it really true? Frankly, what some smart companies brand as AI is simply a clever data analytics algorithm, machine learning or voice and image recognition. Over the past few years, bigger data sets, superior computer processing, together with better automation of manual steps in advertising workflow have contributed to machine learning. So, what is deemed to be AI in advertising could really just be vast knowledge, combined with a sophisticated and pretty user experience design (UX) and buzzwords.

Human intervention is still crucial

As Mark Zuckerberg, Facebook CEO said, Today's AI is good at recognising patterns and bad at what we would call "common sense; without common sense, AI systems can't use knowledge they've learned in one area and easily apply it to another situation. This means they can't effectively react to new problems or situations they haven't seen before, which is so much of what we all do every day and what we call intelligence.

I agree. Common sense and decision making draw on factors beyond data that include senses, awareness, empathy, attitudes, behaviours and thinking. Algorithms cannot influence consumer motivations without first learning the emotions and thought processes of brand guardians. More worryingly, recent studies have shown that machines have picked up on our biasedness resulting in cultural stereotypes and sexist effects. Machines therefore cannot be left to evolve on their own without human intervention. Human intelligence is therefore irreplaceable and will continue to be of great value.

In fact, as another Mark, surnamed Cuban, a billionaire tech investor puts it, the most prized talents in the future are those who can make sense of the data that automation is spitting out and with degrees in studies such as English, Philosophy, and foreign languages or who excel at creative and critical thinking. And these talents certainly do not sound like machines to me.

Will agency roles become obsolete?

While the advertising industry has changed significantly with the onset of digital and continues to evolve, its purpose is still the same. Even as some deliberate about the pros and cons about the implication of ad tech, the basics remain consistent; that is to be able to reach the right audience at the right time in the right place with the right message, and more importantly than ever, at the right price.

Looking ahead, the future of AI in advertising is not to eliminate agencies but rather to improve efficiencies and effectiveness in driving returns on investment (ROI) for clients. We can use AI to capitalise on automated machine learning and big data sets, delivering predictive performance insights such as forecasting the performance of campaigns, further elevating our offerings.

In theory, for marketing teams that demand control, scope and scale, hiring an in-house team for ever-increasing automated digital marketing efforts might sound logical. In practice however, the ever growing and complex digital landscape requires specialised expertise and considerable resources. In fact, these challenges are increasing in intensity with mounting data, content, vendors, platforms and technologies to contend with. In consideration of the time, resources and expenses to recruit and run a well-oiled full service team consisting of digital experts such as traders, social leads, search specialists and data specialists, it is clear why brands are continuing to leverage on the value that external agencies can deliver.

The value of the agency partner remains strong

In recent months we have seen leading companies in the aviation and banking sectors in Asia cutting jobs across departments to focus on leaner and more efficient models. While unfortunate, this appears to be an unavoidable trend that looks set to continue and could potentially affect in-house marketing and digital teams in the future.

In working with agencies, clients can save on the costs of setting up specialist teams, incurring multiple recruitment fees, salaries, onboarding and training costs, office rent etc. This only scratches the surface of the benefits and value a specialist agency provides in terms of cost, productivity, and quality.

The activation of programmatic media requires a unique skill set and range of expert capabilities such as traders, data analysts, partnership specialists and more, to understand and manage the technology and technical platform processes behind programmatic buying. On top of this, the agency is responsible for providing campaign optimisation, reporting, insight and analysis at a granular level. Within our agency for example, a trader works together with an account manager to monitor the ongoing performance of a campaign, making adjustments daily to achieve optimal results for clients.

Given such complexity, the client can maximise the opportunity of an agency relationship, getting the best possible planning and buying, data and technology as well as strategy and insights, without having to carry the resource burden in-house. Essentially, this allows both the brand and the agency to focus on their respective areas of expertise and prioritise results to eventually add value to the bottom line.

Now, do we still need the agency when we have AI? The answer is simple: AI or no AI, creative or media; hiring an agency (and the right one) can only translate to higher ROI.

See the rest here:

Why do we still need the agency when we have AI? - Campaign Asia-Pacific

In Brief Hank from CrashCourse explains in a YouTube video that our perception of humanity might differ in light of the upcoming age of AI. While he does cite many alternate views, he makes it clear that AI is a topic worth discussing.

Crashcourses Hank poses an excellent question at the start of his video. What if your closest friends are a set of extremely advanced robots? How would you know? Hank introduces the Turing Test, an assessment to determine the strength of an artificial intelligence (AI)program. If youspoke toa robot and wereunable to discern a difference between it and a human, then the AI capabilities of its programming would be exceptional.

Hank also spotlights opposing viewpoints which point out that while AImay be able to fool us into believing it is human, it may never truly encapsulate human thought, because humans themselves have yet to fully understand consciousness.

What do you think? If it talks like a human, walks like a human, behaves like a human, and feels like a humanis it human? Do differences in biology matter when the concepts of the mind are the same? Many question whether AI will need rights, and what those rights will look like. It is a question we must turn our attention to.

Read the original here:

Should We Make An AI Kill Switch or Give Robots Rights? - Futurism

Ai (Hebrew: h-y "heap of ruins"; Douay-Rheims: Hai) was a Canaanite royal city. According to the Book of Joshua in the Hebrew Bible, it was conquered by the Israelites on their second attempt. The ruins of the city are popularly thought to be in the modern-day archeological site Et-Tell.

According to Genesis, Abraham built an altar between Bethel and Ai.[1]

In the Book of Joshua, chapters 7 and 8, the Israelites attempt to conquer Ai on two occasions. The first, in Joshua 7, fails. The Biblical account portrays the failure as being due to a prior sin of Achan, for which he is stoned to death by the Israelites. On the second attempt, in Joshua 8, Joshua, who is identified by the narrative as the leader of the Israelites, receives instruction from God. God tells them to set up an ambush and Joshua does what God says. An ambush is arranged at the rear of the city on the western side. Joshua is with a group of soldiers that approach the city from the front so the men of Ai, thinking they will have another easy victory, chase Joshua and the fighting men from the entrance of the city to lead the men of Ai away from the city. Then the fighting men to the rear enter the city and set it on fire. When the city is captured, 12,000 men and women are killed, and it is razed to the ground. The king is captured and put on a stake until he is dead. His body is then placed at the city gates and stones are placed on top of his body. The Israelites then burn Ai completely and "made it a permanent heap of ruins."[2] God told them they could take the livestock as plunder and they did so.

Edward Robinson (1794-1863), who identified many biblical sites in the Levant on the basis of local place names and basic topography, suggested that Et-Tell or Khirbet Haijah were likely on philological grounds; he preferred the latter as there were visible ruins at that site.[3] A further point in its favour is the fact that the Hebrew name Ai means more or less the same as the modern Arabic name et-Tell. Albright's identification has been accepted by the majority of the archaeological community, and today et-Tell is widely believed to be one and the same as the biblical Ai.[citation needed]

Up through the 1920s a "positivist" reading of the archeology to date was prevalent -- a belief that archeology would prove, and was proving, the historicity of the Exodus and Conquest narratives that dated the Exodus in 1440 BC and Joshua's conquest of Canaan around 1400 BC.[3]:117 And accordingly, on the basis of excavations in the 1920s the American scholar William Foxwell Albright believed that Et-Tell was Ai.[3]:86

However excavations at Et-Tell in the 1930s found that there was a fortified city there during the Early Bronze Age, between 3100 and 2400 BCE, after which it was destroyed and abandoned;[4] the excavations found no evidence of settlement in the Middle or Late Bronze Ages.[3]:117 These findings, along with excavations at Bethel, posed problems for the dating that Albright and others had proposed, and some scholars including Martin Noth began proposing that the Conquest had never happened but instead was an etiological myth; the name meant "the ruin" and the Conquest story simply explained the already-ancient destruction of the Early Bronze city.[3]:117[5][6] Archeologists also found that the later Iron Age I village appeared with no evidence of initial conquest, and the Iron I settlers seem to have peacefully built their village on the forsaken mound, without meeting resistance.[7]:331-332

There are five main hypotheses about how to explain the biblical story surrounding Ai in light of archaeological evidence. The first is that the story was created later on; Israelites related it to Joshua because of the fame of his great conquest. The second is that there were people of Bethel inhabiting Ai during the time of the biblical story and they were the ones who were invaded. In a third, Albright combined these two theories to present a hypothesis that the story of the Conquest of Bethel, which was only a mile and a half away from Ai, was later transferred to Ai in order to explain the city and why it was in ruins. Support for this can be found in the Bible, the assumption being that the Bible does not mention the actual capture of Bethel, but might speak of it in memory in Judges 1:2226.[8]:80-82

Fourth, Callaway has proposed that the city somehow angered the Egyptians (perhaps by rebelling, and attempting to gain independence), and so they destroyed it as punishment.[9]

Most archaeologists support the identification of Ai with et-Tell. Koert van Bekkum writes that "Et-Tell, identified by most scholars with the city of Ai, was not settled between the Early Bronze and Iron Age I.[10]Bryant Wood has identified it with Khirbet el-Maqatir but this has not gained acceptance.[11][12]

Coordinates: 315501N 351540E / 31.91694N 35.26111E / 31.91694; 35.26111

See the article here:

Ai (Canaan) - Wikipedia

To create a new drug, researchers have to test tens of thousands of compounds to determine how they interact. And thats the easy part; after a substance is found to be effective against a disease, it has to perform well in three different phases of clinical trials and be approved by regulatory bodies.

Its estimated that, on average, one new drug coming to market can take 1,000 people, 12-15 years, and up to $1.6 billion.

There has to be a better wayand now it seems there is.

Last week, researchers published a paper detailing an artificial intelligence system made to help discover new drugs, and significantly shorten the amount of time and money it takes to do so.

The system is called AtomNet, and it comes from San Francisco-based startup AtomWise. The technology aims to streamline the initial phase of drug discovery, which involves analyzing how different molecules interact with one anotherspecifically, scientists need to determine which molecules will bind together and how strongly. They use trial and error and process of elimination to analyze tens of thousands of compounds, both natural and synthetic.

AtomNet takes the legwork out of this process, using deep learning to predict how molecules will behave and how likely they are to bind together. The software teaches itself about molecular interaction by identifying patterns, similar to how AI learns to recognize images.

Remember the 3D models of atoms you made in high school, where you used pipe cleaners and foam balls to represent the connections between protons, neutrons and electrons? AtomNet uses similar digital 3D models of molecules, incorporating data about their structure to predict their bioactivity.

As AtomWise COO Alexander Levy put it, You can take an interaction between a drug and huge biological system and you can decompose that to smaller and smaller interactive groups. If you study enough historical examples of moleculesyou can then make predictions that are extremely accurate yet also extremely fast.

Fast may even be an understatement; AtomNet can reportedly screen one million compounds in a day, a volume that would take months via traditional methods.

AtomNet cant actually invent a new drug, or even say for sure whether a combination of two molecules will yield an effective drug. What it can do is predict how likely a compound is to work against a certain illness. Researchers then use those predictions to narrow thousands of options down to dozens (or less), focusing their testing where theres more likely to be positive results.

The software has already proven itself by helping create new drugs for two diseases, Ebola and multiple sclerosis. The MS drug has been licensed to a British pharmaceutical company, and the Ebola drug is being submitted to a peer-reviewed journal for additional analysis.

While AtomNet is a promising technology that will make discovering new drugs faster and easier, its worth noting that the future of medicine is also moving towards a proactive rather than reactive approach; rather than solely inventing drugs to cure sick people, focus will shift to carefully monitoring our health and taking necessary steps to keep us from getting sick in the first place.

Last year, the Chan Zuckerberg Initiative donated $3 billion in a pledge to cure all diseases. Its an ambitious and somewhat quixotic goal, but admirable nonetheless. In another example of the movement towards proactive healthcare, the XPRIZE foundation recently awarded $2.5 million for a device meant to facilitate home-based diagnostics and personal health monitoring. Proactive healthcare technology is likely to keep advancing and growing in popularity.

That doesnt mean reactive healthcare shouldnt advance alongside it; fifty or one hundred years from now, people will still be getting sick and will still need medicine to help cure them. AtomNet is the first software of its kind, and it may soon see others following in its footsteps in the effort to apply AI to large-scale challenges.

Image Credit: Shutterstock

See the article here:

Drug Discovery AI Can Do in a Day What Currently Takes Months - Singularity Hub

Maier is head of the governments industrial digitalisation review. Photograph: David Sillitoe/the Guardian

The future has already arrived in a small factory in Worcester, according to the man hired by Theresa May to put Britain at the forefront of the next industrial revolution.

Juergen Maier, the chief executive of Siemens UK, believes new technologies including robotics, artificial intelligence and additive manufacturing, or 3D printing, can deliver greater productivity and create more highly paid jobs.

But failing to crack the next revolution will come at a high price: falling living standards. The work being done in Worcester, and places like it, will be crucial if Britain is to be successful outside the EU, Maier says. The starting gun has been fired in this global race and Britain needs to get ahead.

The beauty of it is, if we get this right, it doesnt just drive productivity, but it also means that youre driving jobs up the value chain, which means that people are getting better paid, so ultimately you raise living standards, the 53-year-old says from the factory floor of Materials Solutions, which is 85% owned by Siemens and boasts big-name clients such as Rolls-Royce.

If you take it as an average, our living standards have hardly risen since the recession. The fundamental reasons are were not exporting enough, and were not driving productivity and output. Unless youre driving productivity, you cant raise wages.

Maier, a firm supporter of the remain campaign in the run-up to the EU referendum, has quickly become the go-to expert on the future of British industry. When he met the Guardian, he was preparing to appear on BBC1s Question Time in Wigan, alongside panellists including the Brexit secretary, David Davis, and the Ukip leader, Paul Nuttall.

His brief as head of the industrial digitalisation review commissioned by the government is to work out how can the UK can better deliver existing technologies, how it can create new industries and, in doing so, whether the UK can generate a net increase in manufacturing jobs despite greater levels of automation. Our gut feeling is we can, but we still need to prove that, Maier says.

The absolute nightmare for me would be that were applying this technology, were displacing jobs as a result of it which will happen but what were not doing at the same time is creating all the jobs in computer science, in data analytics, in software code writing. The good news is we already have a lot of jobs in this area. These industries will create thousands of jobs, software jobs, engineers.

With much at stake, the review is a major undertaking for Maier, but he does have support from a panel of UK business leaders including Sir Charlie Mayfield, the chairman of John Lewis Partnership, and Carolyn Fairbairn, the director general of the CBI. The current plan is to report back with initial recommendations to the business secretary, currently Greg Clark, in late summer.

Maiers ambition for the UK is considerable but so too are the obstacles, not least the uncertainty created by Brexit and strong competition from the likes of Germany and the US. Another key issue will be Britains ability to fill these highly skilled roles in sufficient numbers post-Brexit.

It is not going to be as good as it was in the single market and I think we just have to be more honest about that, Maier says. Im not a moaner or a remoaner, Im saying we have to get to grips with the realities, which are that there are going to be some barriers to trade. And the sooner we accept that the better.

Once weve got over the heat over the elections and this slight hysteria that weve got at the moment, we have to get into a period of calm.

On Materials Solutions factory floor in Worcester, 3D printing machines whirr quietly away, making complex metal parts for industries including automotive, aerospace and motor sports with a speed that would have been unthinkable using traditional manufacturing. Here, welding and forging are replaced with machines that turn 3D CAD models into parts using software, lasers and metal powders.

For Siemens, the firm has delivered breakthrough technology, by 3D printing gas turbine blades. In doing so, the time it takes from the design of a new blade to its production has been reduced from two years to just two months.

Founded and run by Carl Brancher, Materials Solutions employs 23 people and is the perfect example of how Britain can create new, cutting-edge industries, according to Maier.

The Siemens boss believes the prize for the winner in the next industrial revolution is considerable: a thriving manufacturing sector, highly paid, skilled jobs and greater productivity, which would in turn fuel growth and raise living standards for all.

Britain has some advantages, Maier says, including a flexible, skilled workforce and existing infrastructure that fosters innovation such as the UKs Catapult centres, the Alan Turing Institute in London and the Henry Royce Institute in Manchester.

Besides, he says, we have no choice but to win the race to industrial digitalisation: there is too much at stake to fail. We have no option. We have to pull this off and if we dont pull it off, our living standards will drop further.

Read the original here:

Robotics, AI and 3D printing could close UK's productivity gap - The Guardian

Heres a question that will keep future Artificial Intelligence (AI) entrepreneurs up at night: How do you manage a product when the software starts writing itself?

Were not quite there yet, but as we build smarter, more complex software that has elements driven by AI were also making less predictable software. We know that AI will bring more capabilities to software, but it will also make software harder to design and manage since it will sometimes behave in unplanned ways. This is just a phenomenon that comes along with making complex systems. And, thats where we are going with software. This is where complexity theory meets software.

For most of us who have been entrepreneurs, executives, engineers, and product managers in the software industry, we have designed and managed software for decades safely assuming a reasonable level of input-output certainty. Meaning, when we input data, we can easily figure out what the correct output should be. This is because we have been working mostly on simple systems. If you entered A and B into the input, C would come out. If you dont get C, you know you have a defect that needs to be addressed. With simple systems, you can use the same set of test cases over and over again and expect the same outputs over and over again.

Intelligent agents and other dynamic AI-based systems turn this concept on its headas self-learning software adapts its outputs based on inputs from various interactions with other systems and people all the time. Some systems today have gotten pretty complex (especially in the enterprise), but introducing more AI-based algorithms will accelerate complexity beyond where weve been in the past. Well have systems that go from being difficult to decipher why they did something to being indecipherable. And, with intelligent agents, were massively increasing the number of potential inputs (sometimes, the input could be any combination of words in an entire language), which again increases dramatically the number of potential ways to interpret the input and provide a wider array of outputs.

For example, neural nets provide outputs based on inputs, but in between the input and output is the black box of computation. We wont know why exactly the outputs were generated from those particular inputs. And, new training (how the algorithm updates its learning) mean that the outputs may change given the same inputs. So, dynamic updates from a continuously learning piece for software means that there will be layers of learning that happen real-time that will impact outputs in a way that wont be predictable. And, some of these outputs will be fed into other parts of the system, creating additional layers of complexity. We are moving to more complex system design. The term for the new, unexpected things produced by complex systems is called emergence. And, our software will only increase in emergent behaviors as we make them more complex.

This is more of an observation and area of planning than a concern for me. We work with people every day who are unpredictable. No one knows exactly all the reasons people do what they do from moment to moment. Yet, we have found ways to collaborate between humans and get work done. And, for software, well need to think through the issues as we build systems that become more complex. So, based on experience, Ive created some fundamental tips that can help with the above issues as well as other issues when building AI-driven products and AI-based intelligent agents. Note: depending on what you are building, you may need to ignore or alter some of the tips for yourself, based you your particular goals.

Limiting your domain can help limit complexity. So, its a good idea to simplify and focus some things that you have control of, like the domain of expertise of your software. Keep your product constrained into a narrow domain (focused on a logical set of jobs to do for the customer and a logical set of knowledge around an expertise, for example) at first and learn before you expand into other domains.

Every interaction is a chance to learn. Your systems should learn something from all (or almost all) interactions with humans and other systems. Feedback loops are needed for your software to self-correct and learn, and also gives you information to know how to adjust your product and plan for the future. Within your domain, be cognizant of what to optimize for at a high level, but dont over-optimize too soon. Although the AI product can be murky as you explore product feedback loops, you need to choose a more general, large set of capabilities at first and then look for problems that you will be solving for the user. As your user uses the product, your product optimizations can be based on actual customer usage over time.

Sometimes, a human brain is needed to augment the system. Human-in-the-loop refers to the situation where you can have a human complete some tasks to improve a user experience or to figure something out that is too difficult for the system. Designing this in as part of your system will be useful for doing work or validating parts of a process that the system cant do well yet. And, the actions that the human took can feed back into the system to train the system to do the task better for itself in the future. Many companies building AI products use a human-in-the-loop to jump in and do some sort of work as part of their back-end.

Context adds intelligence. (Or, at least the appearance of intelligence.) Were collecting more contextual data than ever, and this context information will be needed for better AI-driven systems across a wide spectrum of industries. For many systems that interact with humans, context will be king. The abilities of intelligent agents will be expanded or constrained based on how much contextual data (location, related data, personalized information, etc.) the application can get. To progress, contextual information will have to be collected directly from the user and any other applications that can be accessed.

Emergent systems require real-time performance evaluation. As we develop systems that operate dynamically, well also need to re-think Q&A. Mainly we need to think about how to augment current Q&A processes. There is more work to be done here, but we will need models for real-time error detection so that we can fail gracefully or have the system jump into another path of action. One way this could be done would be similar to how humans do it by getting feedback from an independent observer. What I mean is an application that constantly observes the production system and looks for abnormal or inaccurate behavior. Once detected, it would give feedback to the main/ production system in order for it to improve and adjust its actions. Sort of like a real-time performance evaluation, except it would be all digital and in real time. I imagine that this application could look similar to virus or spam detection software, where applications can look for a fuzzy determination of normal vs. abnormal behavior.

Expect the unexpected. Humans are unpredictable, and combining unpredictable humans with unpredictable machines exacerbates the issue. Plan for smart failover experiences that can ask for clarity or clearly communicate the confusion to the user. Plan ahead so that the user wont get confused by the dynamic nature of the system.

Use interactive systems to collect good data interactively. When designing inputs to the system via any interface, think about how you can check for the quality and trainability of the data you are collecting. If you are designing an intelligent agent, you can ask the user clarifying questions real-time. If not, you can still build techniques to ensure data quality upon input. There may also be old datasets that could be used to get started with a new customer. Quality will be a factor here as well. Old data sets may not be well maintained and may need to be cleaned up.

Data from users can make the system more valuable, which can help obtain more users and data, which can, in turn, make the system even more valuable. With AI-driven products, information can be collected from all the users on the system (and other systems) to make the system smarter, which in turn makes the system more valuable to attract more users. When you attract more users, data can be collected from them that can feed into the software and so on. This creates a flywheel of data collection and an increasingly intelligent system that builds upon itself. This is a way to create unique value over time. And, it makes it difficult for competitors to catch up as the cycle creates its own momentum.

Give value while collecting data. Balance the collection of data from the user with something useful for the user. The ideal scenario is to provide value while you are learning. Also, if possible, find value in old data that can be loaded into the system through integrations with other systems. Its good to plan for all the great things you can do with data collected in the future, but you have to have some immediate value so that people stick around.

If you are building an intelligent agent, the onboarding never ends. When it comes to intelligent agents, the initial proactive experiences the users have with the agent combined with the ongoing interactions will drive how the user can and will user the agent long-term. So, smart onboarding (introduction to the agent) and ongoing education of the user is key. Humans develop our familiarity with other people through repeated interactions over time. This is how humans will also interact with intelligent agents. If the user and agent havent communicated in a while, then the human may even forget about it altogether. Also, its important to think about how the user will discover what the agent can do. The agent may need to send reminders of new skills it has acquired or even simply provide a visual menu of what it can do. The important thing is to think about how all the capabilities will be presented to the user so that the user understands what it can do and that the user remembers the intelligent agent. The proactive nature of these communications will drive the usage and user expectations needed to do the other things on this list.

In the long term, I predict that intelligent agents will communicate better with humans than humans communicate among themselves. Thats because intelligent agents will have a wider variety of communication methods and input options than humans do. The best path for chat-based or other visual user experiences will usually not be to create a totally text-driven experience. An interface that contains both text elements and visual elements (buttons, etc.) is what we call a hybrid interface and will allow a wide array of input and output options that can be used in the right context to most efficiently communicate. Also, its at this interface point of interaction with users that number 7 (collect quality data) can be enforced. Artful communication with the user is needed to make sure good information is collected that can make the software smarter.

Managing a system requires managing metrics. Metrics are always important for business, especially when you start getting a significant data set from larger numbers of users of your product. Success metrics for AI-driven products will all be slightly different, but they will fall into categories. 1) Quality of data collected that can be used for training, 2) Quality of the modeling in order to generate the right output, 3) AI flywheel growth measurement (for some companies), and 4) customer success metrics (for your particular business, including quality of output to users). As systems get more complex, the right metrics will be needed to ensure you are managing your complex system well.

And, finally, behind many of these thoughts is a common philosophy. We have to start thinking about managing complex computer systems driven by the latest AI-driven capabilities that are capable of emergent behavior. And, that is about managing the parameters, rules, checks, and balances of the system in a way that provides stability for the system. Think about managing an economy. You dont manage an economy (well) by explicitly saying what the prices of all goods and services are. You manage at the higher system level. You set forward a general set of rules (laws) that make sense for that system, and manage a few system level variables (like the federal funds rate). And, the independent agents (in this case, people) will make self-optimizing decisions to set prices by interacting with each other based on their independent needs and wants. Management of complex software systems is similar and will mean designing for good information collection, setting the right parameters, picking the right success metrics for your software, and turning the right knobs at the system level in order to keep the system in the best state of success that you can manage. Therefore part of AI-driven product management is really complex system design and will need more thinking from the perspective of complex systems.

Will Murphy is the VP of Product and Business Development and a cofounder atTalla, an AI-powered customer service company.

More here:

12 tips for designing and managing an AI-driven product - VentureBeat