You will see all three in software marketing, often in the same sentence, often meaning nothing.<\/p>\n
Knowing the difference is useful for one reason: it helps you ask better questions of a vendor.<\/strong> “Which of these does it use, and on what data?” is a question that separates real capability from a label.<\/p>\n Here they are, simply.<\/p>\n They are not three competing things. They are one inside another.<\/p>\n Artificial intelligence<\/strong> is the broadest term. Any software producing useful output from patterns rather than from rules a person wrote out.<\/p>\n Machine learning<\/strong> is a way of doing AI. The system learns from examples rather than being programmed with rules.<\/p>\n Deep learning<\/strong> is a way of doing machine learning, using neural networks with many layers.<\/p>\n Every deep learning system is machine learning. Every machine learning system is AI. Not the reverse.<\/strong><\/p>\n FIGURE 1: ONE INSIDE ANOTHER<\/p>\n Artificial intelligence<\/p>\n Machine learning<\/p>\n Deep learning<\/p>\n The umbrella. It has been used since the 1950s and it covers a lot of things that are not machine learning at all.<\/p>\n An old chess program following hand-written rules was called AI. A rule-based system that decides credit approvals from a decision tree somebody wrote is AI by most definitions.<\/p>\n Which is why “uses AI” tells you almost nothing.<\/strong> It is a category, not a capability.<\/p>\n Here the system learns from examples instead of being told the rules.<\/p>\n How it works, roughly:<\/strong> you show it thousands of past cases with known outcomes. It finds patterns that connect the inputs to the outcomes. Then it applies those patterns to new cases.<\/p>\n What that means practically:<\/strong><\/p>\n It needs data.<\/strong> A lot of it, and clean, with outcomes recorded. This is the single biggest determinant of whether it works.<\/p>\n It handles cases nobody anticipated.<\/strong> That is its advantage over rules.<\/p>\n It is usually right, not always.<\/strong> A statistical answer, not a certain one.<\/p>\n It cannot explain itself well.<\/strong> It can tell you the answer, not a reason you could audit.<\/p>\n Business examples:<\/strong> predicting which invoices will be paid late, scoring which leads will convert, forecasting demand, flagging unusual transactions.<\/p>\n Machine learning using neural networks with many layers. It is what made the last decade’s progress possible.<\/p>\n Where it is genuinely needed:<\/strong> unstructured input. Images, audio, and language.<\/p>\n Reading a scanned invoice in any layout. Understanding a sentence. Recognising a defect in a photograph. These are deep learning problems, and simpler methods do not solve them well.<\/p>\n What it costs:<\/strong> far more training data and far more computing power. And it is the least explainable of the three.<\/p>\n Business examples:<\/strong> document extraction, chatbots, text generation, image recognition.<\/p>\n FIGURE 2: WHICH ONE FOR WHICH PROBLEM<\/p>\n Rules (still AI)<\/p>\n Machine learning<\/p>\n Deep learning<\/p>\n For an ERP, roughly:<\/p>\n Reading scanned documents<\/strong> \u2014 deep learning.<\/p>\n Drafting or summarising text<\/strong> \u2014 deep learning.<\/p>\n Chatbots<\/strong> \u2014 deep learning.<\/p>\n Predicting late payment, scoring leads, forecasting demand<\/strong> \u2014 machine learning, usually not deep.<\/p>\n Approval thresholds, reordering rules, follow-up levels<\/strong> \u2014 rules. Not learning anything, and correct for the job.<\/p>\n That last row matters.<\/strong> A large share of useful business automation is rules, and it is better for those tasks \u2014 predictable, explainable, and exact.<\/p>\n Not “does it use AI?”. These:<\/p>\n What data does it learn from?<\/strong> If the answer is vague, be careful. A model needs your history, and it needs enough of it.<\/p>\n How much of my data does it need before it is useful?<\/strong> For prediction, hundreds of examples is thin and thousands is workable.<\/p>\n Can it explain its answer?<\/strong> If you need to defend the decision to an auditor or a customer, you need a rule, not a model.<\/p>\n Where does the processing happen?<\/strong> On the vendor’s servers, or yours? Ask whether your data is retained and whether it trains their models.<\/p>\n What happens when it is wrong?<\/strong> Every useful business AI has a review step. If it has none, ask why.<\/p>\n FIGURE 3: QUESTIONS THAT SEPARATE CAPABILITY FROM MARKETING<\/p>\n Good answers<\/p>\n Warning signs<\/p>\n Worth saying plainly, because it applies to all three.<\/p>\n They do not understand your business.<\/strong> They recognise patterns. A model that predicts your busiest month has not grasped that you sell umbrellas.<\/p>\n They do not fix bad data.<\/strong> They make bad data reach wrong conclusions faster and more convincingly.<\/p>\n They do not remove accountability.<\/strong> Somebody still signs the accounts and answers to the board.<\/p>\n For deciding what to adopt, the three-way distinction matters less than one question:<\/p>\n How much of my own data does this need, and do I have it?<\/strong><\/p>\n Deep learning on documents<\/strong> \u2014 works out of the box, because it was trained on millions of documents elsewhere. You need none of your own.<\/p>\n Machine learning on your business<\/strong> \u2014 needs your history. Thousands of records with recorded outcomes. Most small businesses do not have this for most things.<\/p>\n Rules<\/strong> \u2014 need no data at all. Just somebody who knows the policy.<\/p>\n That is why document reading pays off immediately for almost everyone, and prediction often does not.<\/strong> It is not that one technology is better. It is that one needs your data and the other does not.<\/p>\n AI<\/strong> is the umbrella term, and it tells you nothing on its own.<\/p>\n Machine learning<\/strong> learns from your examples. It needs volume and clean data.<\/p>\n Deep learning<\/strong> handles documents, images and language. It arrives already trained, which is why it works for you immediately.<\/p>\n Rules are still the right answer<\/strong> for definite policies, and a lot of useful business automation is rules.<\/p>\n Ask what data it needs, whether you have it, and where a person checks the output. Those three questions are worth more than the vocabulary.<\/p>\n Trying to work out which AI claims are real?<\/p>\n Get in touch. We will go through it task by task \u2014 what it does, what data it needs, and whether you actually have that data.<\/p>\n<\/div>","protected":false},"excerpt":{"rendered":" Why bother distinguishing You will see all three in software marketing, often in the same sentence, often meaning nothing. Knowing the difference is useful for one reason: it helps you ask better questions of a vendor. “Which of these does it use, and on what data?” is a question that separates real capability from a […]<\/p>\n","protected":false},"author":3,"featured_media":9136,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[26],"tags":[],"class_list":["post-9135","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai"],"yoast_head":"\nThe nesting<\/h2>\n
\n
\n
\n
Artificial intelligence<\/h2>\n
Machine learning<\/h2>\n
Deep learning<\/h2>\n
\n
\n
\n
Which one your business software uses<\/h2>\n
The questions worth asking<\/h2>\n
\n
\n
What none of them do<\/h2>\n
The practical version<\/h2>\n
The short version<\/h2>\n