Artificial Intelligence - old wine in new bottle...or not ?
Joel Waterman
Active participant in the financial, nuclear energy & digital asset economy
I have to say that every time I see that C3.ai advertisement for AI, I think 'old wine' or 'standard analytics dressed-up as AI'. I can imagine many AI practitioners feeling the same. Anyway, this article is not about C3.ai, a software company, or its marketing campaign - for all I know they do some excellent work. The point is that AI is everywhere and unfortunately everything is branded 'AI' even if it's not. The fact that it's overhyped & over-marketed, does this mean AI is a bubble? Often we cannot 'see' AI at work because the algorithms work in the background. This article attempts to showcase some visible AI. The race is on to deliver the future now.
AI is about creating insights
Alan Turing published his famous work about 'Computing Machinery and Intelligence' in 1954 ! Old wine therefore? AI is about making computers do the work they are best at, finding patterns in large amounts of data and extracting meaning from these. This can be achieved by using plain and simple search, classification and statistical analysis techniques, but it can also be achieved by using neural networks to try and mimic human intelligence and thinking processes. Each have their place and often for the best results a combination of all of the above are necessary. In fact, the most impressive AI results I have seen were obtained by systematically applying multiple techniques on a given dataset and by taking the results of one technique and iterating on it by applying another set of analytics on that data, and so on and so forth. Until an actionable insight was found that could lead to some improvement in a business process or some enhanced understanding of behavior.
Real AI vs Standard Analytics
Using 'business intelligence' software to create a dashboard with regional sales statistics is not AI. Amazon's Alexa in itself is not AI. It's a voice recognition technology that constructs a query over a large database, or most probably, a searchable index and gives back a result. What Amazon does with the data to improve Alexa, how it analyzes voice, how it learns and applies patterns to improve Alexa is AI however. Even then, apart from algorithms, it uses thousands of people worldwide who's job it is to look through data and annotate it to help the computer understand it better (example, they listen to a recording and find that it's about 'Taylor Swift' so they annotate the recording with the words 'musical artist' which a computer can easily find and associate the recording with artists and music).
The difference between statistics and AI is a bit of an academic debate, but generally statistics technologies are used to discover relationships between data whilst AI is used to make predictions, learn from massive amounts of data and the interaction between people and the data and gain deeper insights using things like Machine Learning techniques. As I said above, in the end, depending on what the objective is, different techniques are used on different data sets in order to get the desired outcome.
Making it work takes work (and lots of data)
Gartner's 2021 AI Hype Cycle is about making AI work and that usually involves four initiatives; 1) how to operationalize an AI initiative, 2) what data models and analytics techniques to use, 3) how to remove AI bias, and 4) data, data and more data - specifically how to collect, organize, optimize, store, scale and use it in applications. That takes work.
Data is one of the main stumbling blocks for an effective AI strategy. Often data is siloed, messy, incomplete, under different ownership, not easily accessible or there is simply not enough of it. Many an AI strategy falters at the point of data collection and analysis. Having your data ducks in a row is therefore an essential element to success. Often easier said than done, but very necessary. The other thing that in my own experience holds companies back is the fact that many face a significant challenge in hiring people with the required skills. A report by O'Reilly, a Boston based consulting firm, also confirms this fact (see link below).
What AI is used for
Well known examples are in customer service (virtual agents replacing humans), speech recognition (Siri, Alexa, etc), Image and environment recognition (photo tagging, self-driving vehicles), recommendations (Amazon's 'frequently bought together' or 'consider a similar item'), fraud prevention (for insurance, payments, etc) and multiple examples of trading bots (stocks, investment portfolio composition). Below image is from a McKinsey report from December 2021.
The amount of use cases for AI is more than I can cover here. I have not even mentioned all the amazing work that is done in medical science, robotics and space exploration for example. To put an interactive spin on this I will cover some unusual examples here and things you as the reader can experience and benefit from yourself.
New wine, new methods
Building on my wine theme, this age old industry is innovating using AI technologies. One such example is a company called Vivino. They assembled the world’s largest online wine marketplace, the most downloaded wine app and the Vivino community is made up of approximately fifty million wine drinkers from around the world. The company uses crowd-sourced data to personalize wine recommendations, photo recognition software so you can snap a picture of a label or bottle and use it as your search entry and it has a personalized wine matching service in its wine rating app which will show users how likely a wine will appeal to them using their personal preferences via a ‘match percentage' score.
AI Adoption in numbers
McKinsey did an AI survey in December last year in which they documented the rate of adoption in business globally. It is worth quoting some numbers just to give you a feel for these developments. "Findings from the 2021 survey indicate that AI adoption is continuing its steady rise: 56 percent of all respondents report AI adoption in at least one function, up from 50 percent in 2020. The newest results suggest that AI adoption since last year has increased most at companies headquartered in emerging economies, which includes China, the Middle East and North Africa: 57 percent of respondents report adoption, up from 45 percent in 2020. And across regions, the adoption rate is highest at Indian companies, followed closely by those in Asia–Pacific."
领英推荐
In many ways AI has advanced the most rapidly in consumer technologies whilst it is my impression that, despite the hype, enterprise deployment of AI is progressing but still has a long way to go. In my view this is largely due to the data and talent issues we highlighted above. What has moved significantly is the understanding that this technology is a differentiator and the investment in AI projects has skyrocketed. IDC, in their new Worldwide Artificial Intelligence Spending Guide, forecasts global spending on AI systems will jump from $85.3 billion in 2021 to?more than $204 billion in 2025. The compound annual growth rate (CAGR) for the 2021-2025 period will be 24.5%. Similar data is reported by MarketWatch in their January 2022 report (link below).
Should we fear AI ?
A picture paints a thousand words and this one by The Future of Life Institute sums-up the debate about risks in AI pretty much correctly (click to expand).
What the Brookings Institute Says
AI is so pervasive yet so invisible. It's intensively used but difficult and expensive to do. There is lots of hype and over-use of the term artificial intelligence. At IBM we preferred to use the words Augmented Intelligence, which I still like as it moves the notion away from science fiction and closer to the reality of what the technology promises to do and that is enhancing people's lives, experiences and capabilities rather than replacing them. I like the way the Brookings Institute framed their conclusion. It's a message of hope and expectation about the enormous potential, but also a clear-eyed view on what it takes to get there.
"The world is on the cusp of revolutionizing many sectors through artificial intelligence, but the way?AI systems are developed need to be better understood due to the?major implications these technologies will have for society as a whole."
Sources & Inspirations
SPSS history, https://en.wikipedia.org/wiki/SPSS
IBM's AI website, https://www.ibm.com/it-infrastructure/artificial-intelligence?utm_content=SRCWW&p1=Search&p4=43700068844540368&p5=e&gclsrc=ds
Gartner AI Hype Cycle, https://www.gartner.com/en/articles/the-4-trends-that-prevail-on-the-gartner-hype-cycle-for-ai-2021
Amazon's Alexa and how it learns, https://blogs.scientificamerican.com/observations/how-alexa-learns/ and https://time.com/5568815/amazon-workers-listen-to-alexa/
AI & the wine industry, https://outsideinsight.com/insights/old-world-new-tech-how-the-wine-industry-is-taking-on-ai/ and https://www.forbes.com/sites/bernardmarr/2021/05/07/vivino-choose-your-next-great-wine-with-big-data-and-artificial-intelligence/?sh=56a844a36b94
McKinsey, The State of AI in 2021, https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/global-survey-the-state-of-ai-in-2021
O'Reilly report on the AI Talent gap, https://www.oreilly.com/pub/pr/3314
MarketWatch report, https://www.marketwatch.com/press-release/artificial-intelligence-in-market-share-2021-global-industry-size-growth-trend-demand-top-players-opportunities-and-forecast-to-2027-with-leading-regions-and-countries-data-2022-02-01
Risk & Benefits of AI, https://futureoflife.org/background/benefits-risks-of-artificial-intelligence/#:~:text=The%20AI%20is%20programmed%20to,also%20results%20in%20mass%20casualties.
OECD AI Principles, https://oecd.ai/en/ai-principles
Brookings Institute, https://www.brookings.edu/research/how-artificial-intelligence-is-transforming-the-world/